Open AI Security roles from employers around the world.
Explore the latest roles for AI Security Engineers. Search by role, location, or employer, and filter by region, seniority, employer type, and focus area.
~/AI-SEC-ENG · JOBS.LOG
$aisec jobs sync
>reading board source…
succeeded
>last sync:
08 OCT 2026 09:07 UTC
>sync duration:
1h 07m
>roles found:
241
>employers:
91
$aisec jobs list --open
Open roles in AI security
241 open roles · showing 1–20
Each role is shown as a card on small screens.
Open roles, filtered by the search and filter controls above.
San Francisco, CA, USA; Mountain View, CA, USA; New York, NY, USA
Lead research engineer role at Google DeepMind developing defenses against misuse of frontier AI models. Responsibilities include improving adversarial evaluation techniques, developing out-of-model defenses (classifiers and introspection probes), and enhancing cyber defense capabilities. Apply ML, RL, and cybersecurity expertise to build research prototypes and scalable infrastructure for the cyberattack safety team. Requires PhD, 5+ years cybersecurity experience (red teaming, vulnerability identification), 5+ years agentic tools development, and 2+ years people management experience.
Research Scientist at Google DeepMind focused on advancing AI capabilities for cybersecurity. Develops novel agentic techniques for vulnerability detection and remediation, builds scalable infrastructure for CodeMender's cybersecurity services, and designs evaluation pipelines to measure AI agent performance in vulnerability finding and fixing. Collaborates across DeepMind, Cloud Security, and other organizations on the Cyber Strike initiative. Requires PhD in Computer Science or equivalent, experience with vulnerability research, and strong ML/AI engineering skills.
New York, NY, USA; Mountain View, CA, USA; London, UK
Research Scientist at DeepMind focused on safety and behavior of Gemini models. Responsibilities include developing red and blue teaming methods for frontier GenAI systems, investigating and mitigating safety and security jailbreaks, improving adversarial robustness with focus on high-stakes abuse risks in agentic settings, and delivering scalable mitigations. Requires PhD in Computer Science or equivalent and 2+ years of LLM safety or security experience. Preferred qualifications include agentic workflow development, synthetic data generation, evaluation frameworks, and publication track record.
Develop processes and infrastructure for machine learning red team exercises targeting ML deployments. Plan and execute realistic red team exercises simulating attacker tactics against production ML systems. Design tools, controls, and defensive improvements in cooperation with product security teams. Document findings for technical and executive audiences. Requires 5+ years in security engineering, computer/network security, and threat modeling.
Senior security engineer role at Google DeepMind focused on securing frontier AI serving infrastructure and platform layers. Responsibilities include evaluating and hardening core serving infrastructure (Beyond, Agency, AI Studio APIs), developing threat models for AI-specific attack vectors (prompt injection mitigation, tool-calling hardening, multi-tenant data protection), building secure-by-default libraries and reference architectures, and participating in operational security rotations including on-call incident response. Requires 5+ years in security engineering, platform security, and security operations with coding expertise in systems languages.
Detection Engineering role focused on building and maintaining threat detection programs using agentic AI capabilities. Responsibilities include identifying and closing MITRE ATT&CK detection gaps, orchestrating AI agents across the detection lifecycle, writing behavioral detections using SIEM platforms (Splunk/Elastic), threat hunting, and collaborating with incident response and red teams. Experience required in SOC/incident response, detection engineering, cloud infrastructure (AWS/GCP), and end-to-end agentic detection pipeline development.
Design and build the enterprise AI agent platform at NVIDIA, focusing on runtime safety harness, policy enforcement, agent orchestration, credential management, and observability. Responsibilities include building policy engines for agent action validation, approval gates and kill switches, multi-agent orchestration with secure credential brokering, and evaluation loops to improve agent behavior. Requires 12+ years distributed systems experience, hands-on agent harness experience, and applied security fundamentals including threat modeling, authorization, and secrets management.
Develop and deploy safety solutions for large language models and autonomous AI systems. Focus on LLM security (backdoors, data poisoning, latent behaviors), frontier risks (deception, manipulation), and agentic safety for multi-turn tool-calling systems. Create datasets, evaluation methodologies, and training recipes across safety tracks. Research novel approaches like instruction hierarchy and risk detection. Collaborate cross-functionally to scale LLM security and safety techniques across NVIDIA's frontier models.
GuidePoint Security is hiring an AI Security Engineer to assess and secure enterprise generative AI deployments. The role involves conducting security architecture reviews and assessments of enterprise AI platforms (Claude Enterprise, ChatGPT Enterprise, Copilot) against control frameworks; threat modeling AI workloads for prompt injection, data poisoning, privilege escalation, and supply chain risks; evaluating AI coding tools and development environments for sandbox isolation and secrets access; assessing RAG architectures and vector databases for data exposure; performing shadow AI discovery; and designing secure agentic AI workflows. Requires 3+ years security engineering experience, hands-on enterprise AI platform deployment experience, understanding of LLMs and agentic systems, and ability to produce client-facing security assessments and reference architectures.
Security engineering role at METR, a nonprofit AI risk research organization. Responsibilities span offensive security (red-team exercises, automated AI red teaming), detection and response (AI-powered threat triage systems, detection engineering, incident response), and blue-team infrastructure hardening. The candidate will secure evaluation environments running frontier AI agents, implement detection pipelines and telemetry, conduct incident investigations involving model escapes and infrastructure attacks, and enable safe execution of dangerous AI capability experiments. Requires deep security fundamentals, offensive security experience, AWS expertise, and AI/LLM engineering skills.
Principal-level security architect at NVIDIA responsible for designing, building, and operating core security domains of DGX Cloud AI platform. Owns end-to-end security architecture for GPU cluster isolation, workload identity, infrastructure security, and policy enforcement. Embeds with engineering teams to threat model AI infrastructure, conduct offensive security testing, and engineer classes of risk out of existence. Requires 15+ years in security, infrastructure, and SRE with production-grade coding, experience securing HPC/GPU environments, cloud-native identity systems, and ability to convert offensive security expertise into defensive controls and detections.
Senior Manager leading NVIDIA's AI Safety & Security Engineering team from inception. Responsibilities include hiring and coaching founding engineers across harness/platform engineering and security research; delivering the Find, Validate, and Patch engineering roadmap; and collaborating with evaluation and security leadership. Requires 12+ years software engineering experience with 5+ years managing complex systems teams (distributed infrastructure, ML platforms, security tooling). The role combines AI/ML systems expertise with vulnerability research and security program exposure to build tooling that identifies and patches software vulnerabilities using AI-powered approaches.
Lead strategic agentic AI partnerships for enterprise ISVs, designing and deploying secure multi-agent systems with guardrails, runtime security, and confidential computing. Architect threat-detection and incident-response agent workflows, integrate LLM safety models, and conduct adversarial testing and red-teaming. Build production-grade secure AI infrastructure, create reference architectures, and provide technical guidance on prompt-injection defense, tool-based exfiltration prevention, supply-chain security, and GPU-accelerated safety workflows.
Conduct structured adversarial testing on ByteDance's generative AI models, features, and products to identify vulnerabilities and emerging risks. Investigate jailbreaks, evasions, prompt-based attacks, and other adversarial techniques. Document findings with risk descriptions, reproduction steps, and mitigation recommendations. Partner with cross-functional teams to validate mitigations and develop testing playbooks. Stay updated on emerging adversarial trends including deepfakes and multimodal manipulation.
Washington, DC (with options in San Francisco, Santa Monica, Pittsburgh, Boston; hybrid/remote US-based)
RAND Corporation's Visiting AI Security Resident conducts cutting-edge research on securing AI systems, cyber capabilities, and policy implications. Responsibilities include leading complex projects to develop AI-specific threat models, building software tools for AI cyber capability evaluations, and identifying critical security R&D areas. Work spans technical research, policy analysis, and infrastructure development, with output informing senior government and industry leaders. The three-year appointment requires security engineering or software engineering background, technical management experience, programming proficiency, and ability to communicate across technical and non-technical audiences.
NIST's CAISI Agent Security team seeks research engineers and scientists to conduct cutting-edge security research, assess AI agent vulnerabilities and robustness, and develop agent security best practices and guidelines.
Research Scientist focused on proactive threat mitigation, adversarial machine learning, and agentic security. Drives foundational ML research in model robustness, interpretability, and adversarial resilience. Designs rigorous evaluation benchmarks and stress-testing methodologies for frontier AI systems and multi-agent scenarios. Develops techniques to enhance model resilience against emerging threats including data poisoning, prompt injection, and model backdoors. Collaborates with engineering teams and academic partners to transition research into production solutions. Publishes groundbreaking research in machine learning venues.
Lead foundational ML research in model robustness, continual learning, and interpretability to advance trustworthy AI. Design rigorous evaluation protocols, scenario-based benchmarks, and stress-testing methodologies to assess frontier AI capabilities and multi-agent systems. Develop techniques to enhance model resilience against emerging threats including data poisoning, prompt injection, and model backdoors. Curate datasets and conduct experiments to ensure safety constraint adherence. Collaborate with engineering teams to transition research into production solutions and publish in top ML venues.
Senior security engineer responsible for security construction of ByteDance's overseas cloud products and infrastructure. Conducts security assessments and penetration testing of cloud products, manages vulnerability remediation, and researches attack and defense technologies with AI integration. Requires proficiency in programming languages (Golang, Python, Java, NodeJS), in-depth knowledge of AI security, DevSecOps, or related fields, and hands-on vulnerability discovery and defense capabilities.
Staff-level security researcher focusing on cloud workload security. Research evasion techniques against security sensors, analyze attack paths in cloud environments, identify vulnerabilities in cloud platforms, build detection tools, and create proofs of concept for novel attack techniques. Requires 4+ years in red teaming or offensive security research, deep OS internals knowledge, Python/C++ coding, EDR experience, and hands-on use of agentic AI workflows for detection engineering tasks including reverse engineering and signature writing.
Red team real-world AI-enabled systems (both model and supporting infrastructure) in national security contexts. Develop tactics, techniques, and procedures for attacking AI systems. Write tools in Python, PowerShell, C, and BASH to support red team operations. Requires BS/MS/PhD in computer science or related field with 2–8 years of experience, strong offensive cyber skills (penetration testing, C2 frameworks like Cobalt Strike, reverse engineering, exploit development), TCP/IP mastery, and hands-on Linux/Windows security assessment experience.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)Government
Join Carnegie Mellon's SEI CERT Division to conduct pioneering research in AI security. Responsibilities include developing approaches for analyzing robustness of AI systems, discovering vulnerabilities, reverse engineering AI systems, evaluating defense effectiveness, and influencing AI security strategy. Work with elite professionals and government sponsors to build methodologies addressing emerging threats to AI systems. Requires BS with 8 years, MS with 5 years, or PhD with 2 years in ML, cybersecurity, or related fields, plus practical vulnerability research experience and proficiency with AI/ML frameworks and reverse engineering tools.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Develop machine learning-based prototypes, tools, and systems for AI security applications. Collaborate with researchers to design and execute experimental AI security solutions. Support AI red teaming and adversarial machine learning initiatives. Process and analyze cybersecurity datasets. Apply software engineering best practices to build scalable systems that help solve AI security challenges and influence national AI and cyber security strategy.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Research role at Carnegie Mellon's Software Engineering Institute (CERT Division) focused on AI security. Responsibilities include vulnerability discovery in AI systems, reverse engineering models, evaluating robustness of defenses, analyzing threats to AI systems, and developing methodologies to characterize vulnerabilities. Requires BS in ML, cybersecurity, or related field plus practical experience in vulnerability research and AI/ML implementation. Will develop tools, datasets, and publish findings for government and industry stakeholders.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Join Carnegie Mellon University's Software Engineering Institute (SEI) CERT Division to conduct pioneering AI security research. Develop state-of-the-art approaches for analyzing robustness and vulnerabilities in AI systems. Responsibilities include reverse engineering malicious AI code, vulnerability discovery and assessments, evaluating defense effectiveness, identifying emerging threats, and publishing research findings. Work with government sponsors, industry partners, and policymakers to influence national AI security strategy. Requires BS in ML/cybersecurity with 3 years experience or MS with 1 year experience, practical vulnerability research background, AI/ML implementation skills, and familiarity with reverse engineering tools and Python/C++.
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Red team AI-enabled systems and develop attack tactics against AI platforms in national security contexts. Assess both AI models and underlying hardware/software/network infrastructure. Develop tools and techniques to identify vulnerabilities in AI systems and prepare defenders for real-world threats. Requires offensive cyber expertise combined with AI security knowledge, hands-on penetration testing, exploit development, reverse engineering, and scripting in Python, C, BASH, and PowerShell.
AI Red Teaming
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)Government
Senior researcher role at Carnegie Mellon's Software Engineering Institute focused on advancing state-of-the-art AI security. Responsibilities include developing approaches for analyzing AI system robustness, discovering and assessing vulnerabilities in AI systems, reverse engineering malicious AI code, evaluating defenses and mitigations, and publishing research findings. Requires 5-10 years experience combining AI/ML and cybersecurity expertise, proficiency with ML frameworks and reverse engineering tools, and ability to influence national AI security strategy.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Senior role at CMU's CERT Division developing ML-based prototypes, tools, and systems for AI security applications. Responsibilities include AI red teaming, adversarial ML research, security analysis of AI systems, and translating research concepts into operational capabilities. Work involves processing cybersecurity datasets, designing experimental AI security solutions, and collaborating with researchers to build technologies addressing national AI security strategy.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Conduct research on vulnerabilities and defenses for AI/ML algorithms within Carnegie Mellon's Secure AI Lab. Lead investigations into threat models targeting AI systems, design adversarial attacks and mitigations, and transition research capabilities to government sponsors. Responsibilities include identifying emerging AI vulnerabilities, performing original research on adversarial ML, developing prototype defenses, and mentoring junior researchers.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University – Software Engineering Institute (SEI)AI lab
Staff-level researcher leading efforts to identify and understand security risks in AI systems. Conducts advanced research on vulnerabilities in LLMs, SLMs, and agentic systems including prompt injection, jailbreaking, and data exfiltration attacks. Develops threat models for emerging AI capabilities, designs automated tools to probe AI security mechanisms at scale, and partners with engineering teams to translate research into agentic detection and response capabilities. Mentors researchers and publishes findings internally and externally.
Senior individual contributor designing and implementing security controls for AI platforms, generative AI solutions, AI agents, machine learning environments, and cloud services. Develops reusable security patterns, guardrails, and detection capabilities for enterprise AI adoption. Performs adversarial testing, implements data protection and access controls, conducts security assessments of AI solutions, and provides technical guidance on secure AI adoption. Works embedded with AI Centre of Enablement and Enterprise Technology teams.
Security engineer responsible for automating security enforcement across GPU software releases using AI agents and traditional tooling. Leads adoption of LLM agents in security workflows, developing agentic systems from prototype to production. Owns release security scanning pipeline, conducts threat modeling and vulnerability remediation, and collaborates on security hardening across firmware and user-mode applications. Requires 5+ years security engineering experience in drivers/kernel/firmware, hands-on LLM agent development, and strong Python/C programming.
Security architect for autonomous AI agent platform in government sector. Responsibilities include designing security guardrails for AI agents, threat modeling agentic workflows, building audit trails and explainability mechanisms, implementing data governance controls, and conducting proactive security testing to identify vulnerabilities in AI instruction processing and tool usage. Requires 6–10 years cybersecurity experience with 2–3 years focused on AI/ML or LLM security, with expertise in agentic systems, prompt injection risks, and cloud-native security.
Design, implement, and operate offensive cyber-capability evaluations for frontier AI models and agentic systems. Build realistic evaluation tasks covering vulnerability analysis, exploit development, and multi-step offensive workflows. Execute controlled experiments with AI models, analyze behavior to distinguish genuine capability from evaluation artifacts, and develop cyber ranges with appropriate security controls. Requires offensive security experience combined with AI/model evaluation expertise and programming proficiency.
Lead design and implementation of Microsoft's AI observability, detection, investigation, and hunting platform. Build large-scale distributed data systems for AI security monitoring, attack reconstruction, and threat detection across Microsoft's AI estate. Design detection infrastructure for attack signatures, behavioral anomalies, and model-assisted analysis. Develop high-volume data integrations and correlation pipelines using Spark, Kusto, and streaming platforms. Mentor engineers and drive security architecture reviews. Requires 6+ years distributed systems/security engineering and 2+ years modern AI systems experience (copilots, agents, LLMs, RAG).
San Francisco, CA | New York City, NY | Seattle, WA
Software engineer building safety and oversight mechanisms for AI systems. Responsibilities include developing monitoring systems to detect unwanted model behaviors, building abuse detection infrastructure, surfacing abuse patterns to research teams for model hardening, and creating multi-layered real-time defenses against adversarial inputs and misuse of AI models at scale.
Conduct sophisticated red teaming and vulnerability assessment of generative AI models, including language models and image generation systems. Responsibilities include simulated cyber attacks on AI infrastructure, vulnerability analysis, risk assessment, and development of mitigation strategies. Work with development teams to enhance model security and resilience. Stay current with AI security trends and testing methodologies.
Security Operations engineer focused on AI/ML system security. Responsibilities include developing and maintaining AI-specific incident response playbooks (prompt injection, model poisoning, adversarial attacks), monitoring generative AI systems and LLMs, engineering detection logic for AI security threats, analyzing security implications of new AI/ML platforms before production deployment, reviewing model robustness and adversarial defenses, and participating in AI threat intelligence activities. Contributes to AI governance frameworks, security tooling maintenance, and post-incident analysis for AI-related security events.
Design and build identity services that enable AI agents, enterprise applications, and users to interact securely. Establish authentication, authorization, and credential management for autonomous AI systems operating across enterprise environments. Design identity architectures supporting multi-agent AI platforms with secure agent provisioning, credential lifecycle management, and fine-grained access controls. Implement OAuth 2.0, OIDC, JWT, and Zero Trust principles for AI workflows. Integrate with enterprise identity providers and secure AI tool invocation frameworks. Ensure auditability and governance of AI actions in regulated federal healthcare environments.
Lead design and implementation of secure, scalable AI/ML products across AlphaSense's AI-native platform. Architect security controls for model training, data pipelines, inference environments, and agentic workflows. Identify and mitigate AI-specific attack vectors including prompt injection, data poisoning, model inversion, and model theft. Implement model provenance, integrity, and auditability controls. Embed security throughout the software development lifecycle through threat modeling, secure design reviews, and risk assessments. Define secure development standards and security automation integrated into development workflows.
Design, build, and maintain AI-powered security detections end-to-end using ML models and LLM-based agents to detect sophisticated attacks in EDR telemetry from millions of endpoints. Research attack techniques and TTPs, develop detection hypotheses, implement production detection pipelines that apply machine learning and LLMs to separate intrusions from benign activity, and drive LLM-based agent development for autonomous detection authoring. Requires 5+ years security research/threat hunting experience, deep Windows knowledge, Python development, EDR/XDR familiarity, and hands-on ML/LLM application to security problems.
Staff-level security engineer building AI-powered security automation and tooling within ServiceNow's Product Security R&D function. Develops production security capabilities including AI agent security tools, translates AI security research into engineering, integrates ML models into workflows, and contributes security capabilities across product engineering codebases at enterprise scale.
Santa Clara, CA, US; San Francisco, CA, US; San Jose, CA, US
Principal Security Researcher driving Advanced Threat Prevention detection strategy at Palo Alto Networks. Responsible for vulnerability research, exploit analysis, IPS/IDS detection development, and zero-day response. Leads detection innovation from concept to production, improves rapid response for critical vulnerabilities, and provides technical leadership to researchers. Hands-on focus on translating exploit understanding into production-quality protections, with emphasis on AI/ML-assisted detection workflows and automation to scale threat prevention capabilities.
Senior engineer in GuidePoint Security's Cloud Security Automation and AI Practice. Leads end-to-end delivery of secure AI/ML platform implementations across AWS, Azure, and Google Cloud. Designs and implements secure agentic AI solutions including multi-agent orchestration, AI gateway architectures, and policy-based controls. Architects AI governance policies for enterprise deployments. Develops AI-powered security automation solutions (compliance checks, threat detection agents, remediation workflows). Provides technical oversight, mentors junior engineers, and supports presales activities. Requires 5-7 years experience securing AI/ML workloads in cloud environments with proficiency in Amazon Bedrock, SageMaker, and Azure AI platforms.
Senior Offensive Security Researcher at SentinelOne researching attack evasions, building detection tools, and creating proofs-of-concept for sophisticated attack techniques. Responsibilities include proactively researching latest attack methods, emulating attacks to assess detection engines, leveraging agentic AI workflows for malware analysis and detection engineering, and collaborating with threat intelligence and engineering teams. Requires 4+ years red teaming/offensive research experience, Windows internals expertise, proficiency in Python/C++, malware analysis and reverse engineering skills, and EDR knowledge.
Staff-level AI security specialist focused on securing agentic systems at ServiceNow. Responsibilities include analyzing AI agent security, designing runtime and static controls, threat modeling new AI architectures, red-teaming agents and defenses against prompt injection and tool abuse, building security evaluations and benchmarks, and advising product teams on secure AI design. Combines AI engineering depth with hands-on security control development from prototype to production.
Lead security architecture for NVIDIA DGX Cloud, an AI factory managing hundreds of thousands of GPUs. Set security design standards across tenancy, GPU workload isolation, identity, and supply chain. Embed with engineering teams on real implementation, drive multi-year architecture for common security infrastructure, conduct threat modeling and offensive security testing against AI/ML systems, and lead a small team of Principal Engineers. Build systemic security solutions for large-scale distributed AI platforms, including hardened baselines, admission control, workload identity, supply-chain provenance, and detection tuned to GPU workloads. Serve as technical interface between DGX Cloud and NVIDIA's central security organization.
Design, implement, and manage security across AllianceBernstein's AI infrastructure and products. Assess AI systems for adversarial vulnerabilities including evasion, model inversion, and membership inference attacks. Defend against AI-specific threats such as prompt injection, jailbreaking, and data poisoning. Conduct red-teaming and adversarial testing of AI models and applications before and after deployment. Secure the AI supply chain including model weights, dependencies, and deployment pipelines. Apply cloud security practices (IAM, network segmentation, secrets management) to AI infrastructure. Partner with AI and product teams to build security into development from the start. Support data privacy, governance, and alignment with AI risk frameworks. Translate findings for engineering and non-technical stakeholders.
Design and implement security controls for generative AI systems including LLM applications, RAG pipelines, and agent workflows. Identify and mitigate AI-specific threats such as prompt injection, data exfiltration, and model abuse. Establish guardrails for prompt handling, input/output validation, and safe tool invocation. Secure RAG pipelines through data access controls and retrieval protections. Define model access controls and runtime monitoring. Partner with AI platform teams to integrate security into deployment pipelines. Conduct threat modeling and security assessments for AI systems. Build automated detection and response mechanisms for AI-specific risks.
Build and operate model validation, security scanning, and content safety systems for LLMs in production. Responsibilities include developing pipelines to assess AI models for quality, correctness, and vulnerabilities; implementing model optimization and quantization techniques; building guardrail systems to enforce compliance and mitigate jailbreaking and prompt injection; and designing model evaluation frameworks that measure safety metrics. Requires 5+ years of ML experience, proficiency with PyTorch/TensorFlow, expertise in LLMs and transformer architectures, and familiarity with model safety practices and adversarial robustness.
Design, test, and optimize AI prompts for cybersecurity use cases including vulnerability discovery and attack surface management. Develop AI harnesses and evaluation frameworks for security analysis. Research and implement AI-driven solutions for defensive security operations. Create workflows leveraging AI to analyze security data from SAST, DAST, WAF, and attack surface assessments. Collaborate with engineering and security teams to automate analysis and reduce manual effort.
Senior Security Researcher at Token Security focused on AI Agent and Non-Human Identity security. Conducts deep technical analysis of AI products, LLM frameworks, and agentic tools to identify security gaps and privilege escalation risks. Designs and prototypes security enforcement models for autonomous AI Agents. Performs hands-on offensive and defensive research, develops custom research tools, and publishes technical findings at global security conferences. Collaborates with product and engineering teams to convert research into automated detections and platform features.
Cybersecurity engineer at UK AI Security Institute (AISI) evaluating frontier AI model capabilities in cybersecurity and autonomy. Design cyber ranges and CTF-style challenges to automatically grade AI performance on cybersecurity tasks. Build agentic scaffolding with tools like network packet capture, penetration testing frameworks, and reverse engineering utilities. Design evaluation metrics and interpret results. Conduct infrastructure engineering to ensure robust, scalable evaluation environments. Contribute to research and publications on AI cyber risks. Work directly with frontier labs and UK government partners on pre-deployment testing and capability assessment.
AI Red Teaming
MLSecOps
Security Research
AI Security Institute (AISI), UK Government (DSIT)Government
Design and build evaluations of frontier AI model capabilities in cybersecurity and autonomy using cyber ranges and CTF-style challenges. Build agentic scaffolding to test models equipped with penetration testing frameworks, packet capture utilities, and reverse engineering tools. Conduct pre-deployment testing exercises and validate model behaviors. Work on evaluation infrastructure, metrics design, and research communication for UK government and frontier AI labs.
Software engineer role at the UK Government's AI Security Institute evaluating frontier AI models' capabilities in cybersecurity and autonomy. Responsibilities include building evaluation infrastructure and cyber ranges, developing analysis tooling for model behavior assessment, running pre-deployment testing of frontier models, validating novel model behaviors, and delivering engineering projects. Work directly supports understanding of AI cyber risks, informing safety standards, and developing rigorous evaluation methodologies for advanced AI systems.
AI Red Teaming
MLSecOps
Security Research
AI Security Institute (AISI), UK Government (DSIT)Government
Research Engineer at UK AI Security Institute's Human Influence team, focusing on mitigating AI risks related to human influence and deception. Design RL environments to reduce model ability to deceive users, leverage interpretability methods to identify and mitigate concerning behaviors, and build scalable evaluation pipelines to track AI capabilities. Requires hands-on expertise in LLM post-training, fine-tuning, RL methods, and production ML systems deployment.
Join the UK AI Security Institute's Alignment Red Team to research methods for detecting misalignment in frontier AI systems. Responsibilities include developing automated techniques for finding loss-of-control risks such as research sabotage and reward-seeking behavior, building and running novel alignment evaluations, conducting pre-deployment testing of AI systems, analyzing and reporting findings to frontier labs and governments, contributing to technical publications, and designing software tooling for alignment evaluation. Open to junior, senior, staff, and principal levels.
AI Red Teaming
Security Research
AI Security Institute (AISI), UK Government (DSIT)Government
Join the UK AI Security Institute's Alignment Red Team to research and evaluate misalignment risks in frontier AI systems. Develop techniques for detecting loss-of-control risks such as research sabotage and reward-seeking behavior. Design and build evaluation software, conduct pre-deployment testing with frontier AI companies, and perform threat modeling and incident investigation. Contribute to public research and technical reports. Work autonomously on complex projects combining AI safety research, software engineering, and evaluation methodology to inform government policy and industry safety practices.
Design, develop, and deploy autonomous AI agents integrated with enterprise security infrastructure (SIEM, EDR, XDR). Translate SOC playbooks into agentic pipelines using DAGs and multi-agent routing. Optimize production RAG workflows for security context injection and LLM performance. Implement LLM-specific security controls including prompt injection and data poisoning mitigation. Collaborate with threat analysts on human-in-the-loop escalation mechanisms and secure AI system architecture.
Application Security Consultant performing hands-on assessments of AI/LLM-powered applications and agentic AI systems. Responsibilities include testing for prompt injection, model bypass, excessive agency, tool-calling exploitation, and RAG poisoning attacks mapped to OWASP Top 10 for LLM Applications and Agentic Applications. Builds AI-driven tooling, custom agents, and automation harnesses leveraging LLMs to improve testing coverage and efficiency. Conducts source code reviews, threat modeling, and architecture reviews. Contributes to AI security research and industry publications.
Lead application and AI security strategy across Blink Health's infrastructure. Secure LLM integrations, AI agents, and RAG systems through design reviews and threat modeling. Develop secure patterns for prompt handling, model governance, and AI data protection. Establish secure-by-default practices via SAST/DAST automation, architecture guidance, and DevSecOps maturity. Perform threat modeling and penetration testing for critical systems. Mentor security engineers and influence engineering teams through technical leadership. Leverage AI tooling to improve security operations efficiency.
Lead technical strategy for Google Workspace account security against advanced threats. Drive vulnerability research and adversarial simulation to expose novel account takeover techniques. Design and build advanced agentic systems and simulators to autonomously detect, analyze, and mitigate account security threats. Architect customer-facing agentic tools empowering Workspace and Cloud customers to proactively secure their environments. Investigate high-severity incidents and develop long-term mitigations. Requires 5+ years security engineering experience, coding proficiency, and preferred experience building AI agents and agentic workflows for security automation.
Build autonomous AI security agents that automate alert triage, secure code reviews, threat modeling, and vulnerability assessment for enterprise security teams. Design agent orchestration systems and security controls enabling agents to safely interact with sensitive data in high-trust environments. Conduct security testing of agents and deployment platforms. Research and implement novel security patterns for AI agent systems in production environments.
London, England (remote options: North America, EU, UK)
Research engineer role at Brave focused on securing LLM-powered web agents and chatbots. Design and execute adversarial evaluations to uncover failure modes such as prompt injection and tool abuse. Analyze model behavior to forecast security risks, develop security-focused training techniques, and augment systems with information flow control and privilege management. Partner with engineering and research teams to translate findings into product security improvements and define forward-looking security policies.
Own a major subsystem of a novel exploitability engine that ranks security vulnerabilities by realistic attacker impact. Design and deliver core components including probability modeling, exposure graphs, entity resolution, and calibration loops. Establish AI safety and security guardrails for agentic subsystems. Drive technical direction and partner with security R&D and SecOps to translate customer security problems into production ML solutions.
Founding technical leader for NVIDIA's AI Safety & Security Engineering team. Defines harness architecture and engineering standards for AI-powered vulnerability discovery, validation, and patching systems. Owns technical roadmap for safety-critical AI system design and testing. Bridges exploratory research to production-grade software engineering. Mentors founding engineers and establishes team culture around secure AI development. Requires 18+ years designing complex systems, deep experience with ML pipelines and distributed systems, proven ability to translate research into production, and security engineering background.
Principal ML Engineer setting technical vision and architecture for exploitability-driven security solutions. Defines hardest modeling problems for attack-path probability, threat detection, and vulnerability assessment. Leads AI-safety and security governance for agentic systems in production. Builds scalable, enterprise-grade security products using LLMs, foundation models, and agentic architectures. Sets engineering standards across multiple teams and mentors senior engineers in AI-native security practices.
Develop validation and patch methods for AI-powered vulnerability-finding and patching tools. Establish techniques to confirm vulnerabilities are real and reachable, advance approaches for generating and verifying safe fixes, and define correctness and revalidation standards. Work with LLM-based coding and analysis agents to improve software security. Apply rigorous security judgment to vulnerability classes affecting internal targets, ensuring fixes repair flaws, preserve behavior, and withstand strict revalidation. Maintain disciplined skepticism about AI-assisted tool outputs while leveraging them for analysis and development.
Senior product security engineer at Cohere responsible for securing AI-powered enterprise products. Lead security reviews of architecture and code, threat model agentic AI capabilities, perform hands-on vulnerability testing and PoC development, and build scalable security controls and defaults. Work on emerging risks in AI systems including prompt injection, unsafe tool use, identity/delegation failures, data exposure, and tenant isolation. Partner with engineering teams to develop secure patterns and reduce recurring design vulnerabilities. Requires strong software engineering fundamentals, proficiency in Python/Go/TypeScript, experience leading security reviews for complex systems, and understanding of modern application architecture and untrusted input handling.
Architect and lead development of a novel exploitability-ranking engine that models attack paths and vulnerability risk using probabilistic ML and graph analysis. Own end-to-end system design from evidence ingestion through probability modeling, attack-path search, and validation loops. Establish AI safety, governance, and guardrails for agentic systems. Mentor senior engineers and drive technical direction across architecture, design, and security integration with production validation against incident replay and purple-team testing.
Principal engineer role focused on defining and building core agentic identity and security capabilities for trusted AI agents at NVIDIA. Responsibilities include architecting agent identity security features across cloud and on-premises environments, prototyping solutions for credential brokering and workload identity, building reusable APIs and tooling for secure agent workflows, and applying adversarial thinking to security architecture decisions. Requires 15+ years in software engineering and security with hands-on technical leadership experience.
Principal engineer role focused on designing and building core agentic identity and security capabilities at NVIDIA. Responsibilities include architecting agent identity features for cloud and hybrid environments, prototyping secure credential brokering and token exchange mechanisms, building reusable security APIs and tools, and evaluating emerging agentic identity standards. Requires 15+ years in security/identity engineering with hands-on technical leadership, strong offensive security and red teaming background, and expertise in AI-specific security risks including prompt injection and tool misuse.
Build production ML components for an exploitability engine that ranks security work by exploitability. Design and operate core systems for evidence ingestion, entity resolution, graph construction, and probability scoring. Apply AI safety and guardrails practices while partnering with security teams to solve vulnerability management and threat detection problems. Requires strong software engineering, hands-on ML/LLM application experience, and security domain interest.
Remote-Friendly, United States; San Francisco, CA; Seattle, WA; New York City, NY; Washington, DC
Conduct red team and purple team engagements against Anthropic's cloud infrastructure, endpoints, and bare metal deployments. Perform penetration testing on high-value deployments, develop AI-assisted security testing tooling, and contribute to AI-specific attack scenarios. Requires 5+ years of hands-on red teaming and offensive security operations, with deep expertise in cloud security, Linux/macOS security, or Kubernetes. Track record of discovering novel attack vectors and chaining vulnerabilities creatively. Experience with AI/ML systems and agentic workflow testing preferred. Present findings to technical and executive stakeholders.
Design and build scalable ML infrastructure to support real-time safety deployments for Claude's ML-based safety classifiers and defenses. Develop monitoring and observability tools for classifier performance and data quality in safety-critical applications. Collaborate with research teams to productionize safety research into robust, scalable systems. Optimize inference latency and throughput for real-time safety evaluations. Implement automated testing, deployment, and rollback systems for ML models in production safety applications. Partner with Safeguards, Security, and Alignment teams to deliver infrastructure meeting safety and production needs.
Set platform-wide security strategy and architecture for ServiceNow's agentic AI, LLM, and model systems. Lead threat modeling for cross-agent autonomy and novel GenAI attack surfaces, define secure-by-design standards for agentic systems, and establish architectural patterns for securing model/RAG/data pipelines. Mentor engineering teams on AI-specific security and represent ServiceNow's AI security posture externally. Requires 10+ years software engineering with 7+ years security focus, demonstrated track record building AI red-teaming programs, LLM/agentic threat modeling at scale, and hands-on experience securing data pipelines.
Senior security engineer role at OutSystems focused on securing AI-powered and agentic capabilities across the platform. Responsibilities include leading security initiatives across the SDLC, threat modeling for agentic systems, security assessments of applications and APIs, developing secure-by-default patterns, operating and improving security tooling, and mentoring junior engineers. The role emphasizes security by design for autonomous systems and requires hands-on experience with application security, cloud-native environments, AWS, Kubernetes, and understanding of AI-enabled systems.
Design and implement ML-driven security systems for identity risk assessment, anomalous access detection, malware classification, and sensitive data discovery. Architect high-performance inference pipelines and reasoning systems for real-time security decisions. Apply machine learning to access control, behavioral biometrics, and identity confidence scoring. Develop attack detection and threat classification models with focus on false-positive reduction. Lead technical initiatives combining ML research, systems engineering, and security expertise. Build ML platforms, feature engineering frameworks, and model management infrastructure for production security-critical environments.
Research Engineer at Anthropic focused on securing AI systems through reinforcement learning and agentic integrations. Responsibilities include designing RL environments for incident response and security analysis, developing agentic systems for autonomous investigation and remediation, and conducting security evaluations. Requires machine learning expertise combined with cybersecurity domain knowledge. Role involves both research innovation and production engineering, collaborating across security, safety, and research teams. Ideal candidates have ML research or security professional backgrounds with software engineering skills and interest in defensive cybersecurity applications.
Research Engineer role at Anthropic focused on advancing AI security through reinforcement learning. Responsibilities include developing agentic integrations for autonomous security investigation and remediation, designing RL environments for incident response and vulnerability analysis, conducting security experiments and evaluations, and delivering solutions into production training runs. Requires ML experience combined with cybersecurity domain expertise, strong software engineering skills, and ability to balance research exploration with implementation. Experience in security engineering, detection and response, or applied defensive work is preferred.
Principal Security Researcher conducting cutting-edge security research on AI-powered DevSecOps platform. Responsibilities include developing novel testing methodologies for AI agent security, conducting hands-on penetration testing, validating vulnerabilities through proof-of-concept exploits, assessing AI attack vectors (prompt injection, agent manipulation, workflow exploitation), and leading security research into GitLab's AI and agentic surfaces. Will define security requirements for AI systems, build agent-assisted vulnerability discovery tooling, and mentor team members. Requires 10+ years security research/offensive security experience, strong coding proficiency, AI framework knowledge, and deep understanding of AI attack vectors.
Remote (Canada, Israel, United Kingdom, United States)
Staff Security Researcher conducting cutting-edge security research on GitLab's AI-powered DevSecOps capabilities, including Duo Agent Platform and GitLab Duo Chat. Responsibilities include identifying and validating vulnerabilities in AI and agentic surfaces, developing novel testing methodologies for AI agent security, conducting penetration testing, validating vulnerabilities through hands-on exploit development, and building automation for agent-assisted vulnerability discovery. Requires 7+ years in security research/penetration testing, expertise in multiple technical areas, proficiency in Ruby/Go/Python/TypeScript/Rust, and understanding of AI attack vectors including prompt injection and agent manipulation.
Staff-level security engineering role focused on threat detection and response using AI agents and agentic workflows. Responsibilities include designing bespoke and standard security tooling, threat hunting against AI-driven threats, incident response automation, threat modeling, and security architecture. Requires 3+ years production cloud security experience, expertise in AI agent design, MITRE ATLAS threat framework knowledge, SIEM/SOAR platforms, and cloud infrastructure (AWS/GCP). Position emphasizes building scalable detection systems, automating security processes, and solving complex threats at enterprise scale.
Staff+ SRE role designing, building, and operating production infrastructure for Claude's safety systems. Owns deployment of safety classifiers and safeguards across model launches (1P, AWS Bedrock, GCP Vertex). Responsibilities include launch verification, canary rollouts, configuration drift detection, automation of safety validation pipelines, safeguards registry maintenance, and on-call incident response. Requires deep production change management experience, high-stakes release leadership, hands-on cloud platform operations (AWS/GCP), and Python proficiency.
Fort Meade - Annapolis Junction, Maryland, United States of America
Lead technical teams at MITRE's Trustworthy & Secure AI Department performing AI adversarial threat analysis and defense research. Assess threats to AI systems, develop mitigations against adversarial tactics, and integrate secure AI solutions into government systems. Requires 8+ years AI experience, hands-on expertise in agentic AI, deep learning, generative AI, and demonstrated AI red teaming or adversarial AI experience. Provide thought leadership on AI capabilities applied to cyber challenges and mentor technical staff. Active TS/SCI clearance required.
Lead and manage a high-performing security engineering team at Cohere focused on securing AI platforms and enterprise AI systems. Responsibilities include designing and implementing vulnerability management, SAST/DAST programs, leading penetration testing of web applications and agentic AI systems, managing bug bounty programs, and embedding security practices throughout the SDLC. Requires 8+ years of application security experience with expertise in vulnerability management, cloud security, and secure development practices.
Senior Security Engineer role focused on strengthening security and resilience of AI/ML ecosystems. Responsibilities include identifying and assessing security risks in ML models and generative AI applications, designing and implementing security controls, building security frameworks for safe AI adoption at scale, integrating security into the AI development lifecycle, automating security assessments, and contributing to AI security practices evolution. Required expertise in security frameworks (OWASP), Python programming, CI/CD pipelines, security tools (SAST, DAST, SCA), and high-scalability environments. Nice-to-have skills include experience with AI frameworks (PyTorch, TensorFlow), HuggingFace, and cloud infrastructure.
Research and testing role on the AI Security Institute's Control Red Team, focused on stress-testing control monitors designed to detect misaligned behaviour in advanced AI systems. Responsibilities span two tracks: (1) research on measuring control measure efficacy through ML experiments, RL optimization, and adversarial attack design; (2) testing frontier labs' monitors through threat modelling, breaking monitors/sandboxes, security analysis, and producing decision-relevant reports. Includes building tooling, experimental pipelines, and infrastructure for model training and serving at scale. Heavy use of LLMs to automate attack and evaluation loops.
Lead the Systems team within Anthropic's Safeguards group. Own the real-time infrastructure that classifies tokens produced by Claude models and determines safety. Manage engineers building critical infrastructure for token classification, safety enforcement decisions, and cost-effective deployment. Partner with Safeguards and core infrastructure teams to ensure fast, available, correct systems that keep pace with evolving safety requirements. Requires team management experience and deep technical background in high-availability distributed systems.
Join AISI's Misuse Red Team to conduct adversarial testing of frontier LLM safeguards. Research Scientists lead technical direction on novel attack and defense research; Research Engineers build systems enabling large-scale experimentation. Responsibilities include developing and analyzing attacks on LLM-based systems, researching novel attack vectors, creating automated attack tooling, and publishing findings shared with frontier AI developers and governments. Work spans jailbreaking algorithms, poisoning attacks, agent misuse evaluation, and fine-tuning API defenses.
Security Software Engineer supporting the operations and development of NVIDIA's AI-powered vulnerability management platform. Responsibilities include maintaining production scanning fleets powered by LLM agents, building full-stack features for the vulnerability hub, triaging and fixing issues, operating GitOps deployment pipelines, and collaborating on LLM agent harnesses that power automated security scanning across repositories and web applications. Requires Python, SQL, Linux fundamentals, and growing security expertise.
Build and improve LLM-based red team agents for autonomous reconnaissance, exploitation, and evidence capture. Design multi-agent orchestration systems, encode operator tradecraft into prompts and tools, develop exploit-confirmation harnesses to reduce false positives, and engineer safety controls for autonomous offensive capability. Requires 12+ years security engineering with 4+ years offensive security (penetration testing, red teaming, exploit development), strong Python software engineering, practical LLM/agent framework experience, and hands-on exploitation skill in web applications, cloud/Kubernetes, or systems/binary domains.
San Jose, CA; Austin, TX; Boston, MA; Dublin, Ireland
Security researcher at Vectra AI focused on developing and validating AI-driven threat detection capabilities. Responsibilities include researching network-based threats, collaborating with Data Science teams to develop detection models, replicating attacker techniques for detection validation, and testing efficacy of AI detection systems. Requires 5+ years of attack/penetration testing, security research, malware analysis, or incident response experience, plus proficiency with attack frameworks and network forensics tools.
Individual contributor role focused on guardrails and AI red-teaming for foundation models. Responsibilities include identifying edge-case vulnerabilities, designing and training ML models and small language models in security contexts, working hands-on across model, data pipeline, and API layers, and providing reproducible findings with remediation guidance and retesting. Fully remote team across Canada.
Design and execute red-teaming campaigns against AI monitors and coding agents, identifying novel attack surfaces and failure modes. Build automated adversarial testing pipelines that scale red-teaming efforts. Collaborate with control researchers to develop iterative adversarial games and recommendations for monitor improvement. Track literature on agent failure modes and monitor evasion. Produce publications and internal reports on campaign findings. Requires 2+ years in offensive security, adversarial ML, or AI red-teaming; strong experience with coding agents and structured adversarial testing; Python proficiency; and demonstrated ability to work independently on open-ended adversarial problems.
Design threat models and control protocols for coding agents at risk of misalignment or compromise. Maintain failure mode libraries, develop attack trajectories for monitor backtesting, adjudicate flagged agent behavior, and red-team Watcher's security monitors. Requires 5+ years hands-on security engineering, threat modeling expertise, code/infrastructure analysis, and ability to reason about insider-risk analogues in agentic AI systems.
Remote-Friendly, United States; San Francisco, CA; Washington, DC
Design and run evaluations measuring cyber-relevant capabilities and safeguard robustness in AI models. Execute per-release safeguard-robustness testing, analyze jailbreaks and prompt bypasses, design detection probes for cyber misuse, and build layered abuse-detection architecture. Requires hands-on cybersecurity experience including CTF participation, vulnerability research, or exploit development, combined with ML evaluation framework expertise.
Washington, DC / San Francisco, CA / Santa Monica, CA / Pittsburgh, PA / Boston, MA (hybrid/remote options)
Lead multidisciplinary teams at RAND's Center on AI, Security and Technology to develop and execute systems that evaluate AI model vulnerabilities across attack lifecycles (initial access, lateral movement, defense evasion). Build benchmarks for autonomous AI operations in adversarial environments, including CTF frameworks and attack graph reasoning assessments. Design rigorous threat models and evaluation frameworks to assess AI stealth and adversarial capabilities. Communicate technical findings to senior government and industry leaders to inform responsible AI policy. Manage research budgets, complex technical projects, and lead cross-functional teams of engineers and researchers.
Design and implement control protocols and monitoring systems for securing AI coding agents. Conduct empirical research to test monitor effectiveness across failure modes, build evaluation frameworks and datasets, fine-tune models for production monitoring, and incorporate adversarial red-teaming findings into detection systems. Requires 2+ years empirical LLM/AI research experience, strong coding agent expertise, LLM-as-a-judge experience, and Python proficiency. Work closely with research and product engineering teams on real-world agent deployments.
Security Researcher at Microsoft's AI Red Team responsible for discovering and exploiting GenAI security vulnerabilities in frontier and foundational models across Microsoft's AI portfolio (Bing Copilot, Security Copilot, Github Copilot, Office Copilot, Windows Copilot). Develop red teaming methodologies to assess AI systems and models for security and safety failures. Work with adversarial ML researchers, safety experts, and product development teams to identify risks before launch. Requires bachelor's degree in CS/engineering with 2+ years related experience or equivalent, plus preferred experience in penetration testing, threat analysis, cybersecurity, or vulnerability research. Ability to test agents and familiarity with GenAI systems preferred.
Senior AI Security Researcher at NVIDIA focused on developing methods, tools, and evaluations to test, attack, and defend frontier AI systems, agentic applications, and AI-enabled security automation. Responsibilities include discovering novel failure modes in AI models and autonomous agents, building rigorous evaluation harnesses, prototyping adversarial and defensive techniques, exploring LLM security, agent security, adversarial testing, model evaluation, and cyber-defense automation, and translating research into practical mitigations and secure-by-design recommendations for engineering teams.
Design, deploy, and evaluate AI-powered cybersecurity solutions within virtualized environments. Fine-tune NVIDIA Nemotron models for threat detection and automated response tasks. Build CI/CD pipelines for security product validation and benchmarking. Curate datasets for security analytics. Work at the intersection of AI development and cybersecurity with offensive security researchers to advance AI-powered defense capabilities.
McLean, Virginia; Bedford, Massachusetts, United States
MITRE's Trustworthy & Secure AI Department seeks an AI Security Engineer to assess threats to AI systems, develop defenses against adversarial tactics, and integrate secure AI solutions into government systems. Responsibilities include adversarial threat analysis, vulnerability characterization in national security systems, MITRE ATLAS framework application, and capability innovation for cyber challenges. Requires Bachelor's degree plus 2 years AI experience (or advanced degree), advanced AI/ML theory knowledge, hands-on expertise in agentic AI or generative AI, Python proficiency, and ability to obtain TS/SCI clearance.
Remote (London preferred, North America/EU/UK accepted)
Design and execute adversarial evaluations to uncover failure modes in LLM-powered web agents and chatbots, including prompt injection and tool abuse. Analyze model behavior to understand security risks, develop security-focused training techniques, and augment systems with information flow control and privilege principles. Partner with engineers and researchers to translate findings into product improvements and measurable security gains. Help define forward-looking security strategy for building and releasing AI agents responsibly.
MITRE's Trustworthy & Secure AI Department seeks a Senior Adversarial AI Engineer to perform AI adversarial threat analysis, identify vulnerabilities in national security AI systems, and develop defenses against adversarial tactics. Responsibilities include threat assessment, vulnerability characterization, intelligence analysis and reporting, and technical subject matter expertise. Requires 5+ years AI experience with hands-on expertise in Generative AI, AI Security, Agentic AI, Testing & Evaluation, or Applied AI. Python and AI technical stack proficiency required. Active TS/SCI clearance mandatory.
Lead AI Red Team within Threat Operations at Amazon. Build and manage security engineers conducting offensive security research targeting AI systems, training pipelines, inference systems, and model architectures. Establish strategic vision for AI offensive security research, oversee sophisticated red team operations across AI portfolio, develop scalable offensive security automation and AI-augmented testing tools, and drive cross-organizational security initiatives to assess vulnerabilities introduced by growing AI reliance. Translate technical findings into strategic recommendations for leadership.
AI Security Researcher at Apollo Research. Responsible for identifying, researching, and remediating conventional and novel AI agent-related threats. Red-team internal software, infrastructure, and AI agent access controls. Design solutions for emerging AI security threats, track adversary tactics relevant to AI systems, and own findings through deployed fixes. Build detections, controls, and durable security infrastructure. Requires 5+ years hands-on security experience including offensive security, threat modeling, and red teaming. Engineering mindset essential for translating security findings into automated controls and system-level solutions.
Lead product engineering team for Watcher, a coding agent security product deployed in production monitoring billions of agent tokens monthly. Ship code in Python/TypeScript for agent security monitoring and control systems. Manage product engineers and research scientists. Coordinate delivery of security monitors, threat detection research outputs, and enterprise agent deployments. Set technical standards for secure agent architectures and incident response.
Forward Deployed Engineer owning end-to-end deployment and integration of Watcher, a production coding agent security platform. Responsibilities span deployment engineering (40%), including cloud/on-prem rollouts and agent/SIEM integrations; pre-sales and technical evaluation (30%), running POCs and security assessments with prospects; and white-glove support (30%), providing named technical contact for key customers and driving product feedback. Strong engineering foundation (3+ years production software) required, plus excellent customer communication, high agency, and comfort with enterprise security environments. Experience with coding agents, SIEM tooling, and pre-sales a plus. Salary: $222k–$290k (SF) or £149k–£195k (London).
Full-stack engineer building Watcher, a production security platform for coding agents. Responsibilities span agent-side hooks and CLI tooling, log ingestion and real-time monitoring pipelines, enterprise monitoring dashboards, and security controls (tenant isolation, encryption, access management). Own features end-to-end from requirements through production deployment. Design and operate backend systems processing high-volume agent logs with robust error handling and observability. Build integrations with enterprise security stacks and support flexible deployment models (cloud, on-prem, local). 4+ years production software experience required; full-stack capability (Python, TypeScript) essential. Strong product sense and startup-pace execution mindset.
Lead the development and implementation of Bupa's AI security architecture, standards, and governance frameworks. Define enterprise-wide AI security controls, design threat models for AI initiatives, establish risk assessment criteria, and provide strategic advisory support for secure AI adoption across the organization. Collaborate with technology, security, risk, and governance functions to balance innovation with security in a regulated healthcare environment.
Senior Security Engineer at Apollo Research focused on designing and deploying security controls for AI agent infrastructure. Responsibilities include building core security systems (IAM, cloud security, endpoint management, incident response), establishing security practices across the organization, evaluating security tooling, and leading major security projects. The role requires 5+ years of hands-on security experience with application/cloud/product security expertise, ability to threat-model novel AI agent risks, and engineering mindset to build custom solutions. Will work on zero-trust migrations, agentic SOC automation, sandbox hardening, and AI agent pentesting tooling.
Lead security engineering organization protecting an AI-first, cloud-native SaaS platform. Own application/product security, infrastructure/cloud security, and vulnerability management across distributed US and India teams. Responsibilities include AI product security (threat modeling, secure patterns for models and agents, LLM/RAG security), vulnerability operations (discovery, prioritization, remediation), cloud/infrastructure security (AWS, infrastructure as code, containers), and supply chain security. Requires 8+ years in security with 3+ years leading distributed teams, practical AI/ML security knowledge, and experience with secure software development lifecycle, application security testing, and cloud-native platforms.
Senior cloud security engineer responsible for designing and implementing autonomous AI agents for security workflows and incident response automation. Develops systems for automated vulnerability detection and remediation in enterprise cloud infrastructure. Requires hands-on expertise in AI security, adversarial testing, prompt injection defense, AI model security, and ML supply-chain security. Designs secure cloud architectures, implements security automation with IaC tools, and leads cloud security teams across AWS and Azure environments.
Lead distributed security engineering organization protecting 6sense's AI-enabled, cloud-native SaaS platform. Owns vulnerability operations, infrastructure security, and application/product security with strategic responsibility for AI-specific risks including prompt injection, agent access control, model integrity, and data pipeline security. Combines people leadership of cross-region teams with hands-on technical engagement in secure development lifecycle, threat modeling, architecture review, cloud security posture, and vulnerability governance. Reports to CISO.
Build and maintain AI agents and infrastructure supporting NVIDIA's security organization, including agent-enabled workflows for certifications, risk, and compliance. Design and implement MCP integrations, ETL pipelines, and data infrastructure serving security agents. Requires 8+ years automation/data engineering experience, proficiency in Python/Go/C++, production AI/agent experience, and AWS/Terraform/Airflow expertise. Background in security or cybersecurity valued.
Individual contributor designing and executing hands-on adversarial testing across model, application, agentic, and pipeline layers. Responsibilities include multi-turn jailbreaks, guardrail bypass, prompt injection, agent/tool misuse, dangerous-capability evaluation, and data poisoning. Deep-dive investigation of edge-case vulnerabilities, severity ranking per OWASP/NIST/MITRE/EU AI Act, and remediation validation. Requires expert Python, PyTorch/TensorFlow/HuggingFace, ML fine-tuning (LoRA/QLoRA/PEFT), adversarial attack design, and MLOps security knowledge.
Senior-level AI red team engineer role at Carnegie Mellon University's CERT Threat Analysis Directorate. Conducts adversary emulation exercises and red teaming against real-world AI-enabled systems in national security contexts. Responsibilities include developing offensive tactics and techniques for attacking AI systems, writing exploitation tools in Python, C, BASH, and PowerShell, reverse engineering AI platform components, and assessing security across hardware, software, and network layers. Requires 8+ years offensive security experience with penetration testing, red teaming, or exploit development background, proficiency with C&C frameworks, reverse engineering tools, and TCP/IP expertise. Position involves 25% travel and requires Department of War security clearance.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Lead AI security strategy development at Siemens DI SW, reporting to the Chief Product Solution & Security Officer. Design and deploy security measures for AI models, data, and infrastructure. Develop security frameworks for AI systems and datasets. Conduct risk assessments and threat modeling to identify vulnerabilities. Lead research into adversarial attack detection and AI security defenses. Analyze AI security incidents and conduct forensic investigations. Develop tools to monitor AI security risks and detect anomalies in model behavior. Collaborate with AI researchers, software engineers, and security teams. Ensure compliance with NIST AI Risk Management frameworks and secure AI development guidelines.
AI Red Teaming
Detection & Response
LLM Security
Siemens Digital Industries SoftwareSecurity vendor
Red team AI-enabled systems and related infrastructure in support of national security objectives. Develop tactics, techniques, and procedures for attacking AI platforms. Write exploitation tools and conduct adversary emulation exercises. Requires bachelor's degree in computer science or related field with 3+ years relevant experience (or MS with 1+ years), proven penetration testing and red teaming experience, proficiency in Python/C/BASH/PowerShell, reverse engineering tool expertise, and relevant security certifications (OSCP, CPTS, FORGE/RIOT, eJPT, CBBH, BSCP, PNPT, or GRTP).
Lead engineer for NVIDIA's Agent Policy Fabric core platform, responsible for building runtime policy verification, signed policy bundle handling, trust-root management, policy projection, conformance testing, and cross-runtime integration for governed agentic systems. Design and harden authorization APIs, cryptographic verification pipelines, and fail-closed security defaults. Collaborate with runtime teams on public interfaces for policy enforcement, runtime attestation, and audit. Define versioning, latency budgets, and architecture specifications for product security and enterprise partner adoption.
NVIDIA seeks an AI red teamer to reduce risk and threats in networking AI products. Responsibilities include hands-on safety and security research on AI/networking systems, developing tools to expose model and infrastructure vulnerabilities, defining AI development security standards, and partnering across teams. Requires 5+ years in AI safety/security and offensive cybersecurity, expertise in LLMs/generative AI/agents/RAG, Python programming, and MLOps knowledge (Docker, Kubernetes). Strong candidates have software product design experience and familiarity with PyTorch/TensorRT.
Design, build, and deploy AI agent systems for enterprise security workflows including PQC transition, threat modeling, and automated mitigation. Architect LLM and RAG integrations with security data sources. Own full lifecycle from architecture through productization and operational deployment. Requires 5+ years experience with agentic systems, LLM production deployments, strong cryptography and secure design foundation, and Python system design expertise.
Red team AI-enabled systems and related infrastructure to identify vulnerabilities and develop defensive countermeasures. Conduct adversary emulation exercises against real-world AI platforms used in national security contexts. Develop tactics, techniques, and procedures for attacking AI systems. Write offensive tools in Python, PowerShell, C, and BASH. Perform penetration testing and vulnerability analysis on AI platforms across hardware, software, and network layers. Reverse engineer and analyze code to identify security weaknesses. Engage in field operations and present findings to mission partners.
Red team AI-enabled systems used in national-security contexts, combining offensive cyber tradecraft with AI security expertise. Develop attack tactics, techniques, and procedures against AI systems and their supporting infrastructure. Conduct adversary emulation exercises, write exploitation tools, and prepare defenders against real-world AI threats. Requires proficiency in penetration testing, exploit development, reverse engineering, C2 frameworks, and programming in Python, C, and BASH.
AI Red Teaming
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)Government
Red team AI-enabled systems used in national security contexts, combining offensive cyber tradecraft with AI security expertise. Develop tactics, techniques, and procedures for attacking AI systems and supporting infrastructure. Write tools in Python, PowerShell, C, and BASH. Conduct adversary emulation exercises, reverse engineering, vulnerability analysis, and real-world testing of AI platforms. Requires penetration testing, exploit development, and command-and-control framework experience.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Build trusted AI agents that augment security analyst judgment and automate detection, investigation, and response workflows. Develop context-aware agents analyzing security telemetry, design security-specific evaluations and benchmarks, optimize models and agents using production traces and analyst feedback, and take capabilities from experimentation through evaluation to production deployment. Provide technical direction for agentic AI initiatives and influence architecture decisions.
Senior Data Scientist in NVIDIA's Security and Networking Architecture group developing agentic AI systems for automated threat analysis and response workflows. Responsibilities include building and optimizing generative models with RAG and tool-augmented reasoning, fine-tuning models for production security systems, developing ML algorithms across multiple data types, creating domain-specific security datasets, and benchmarking system performance. Requires MS/PhD, 5+ years production ML/DL experience, expertise with PyTorch/TensorFlow, agentic AI frameworks, LLM fine-tuning, and inference optimization. Security and networking background preferred.
Lead and execute advanced AI red teaming engagements against large language models, RAG systems, and AI agents. Conduct multi-stage adversarial assessments including prompt injection, jailbreaking, tool abuse, model manipulation, and data exfiltration testing. Perform comprehensive penetration testing across web applications, APIs, mobile, cloud, and infrastructure. Execute secure source code reviews and red team exercises. Produce technical reports with actionable remediation recommendations and advise clients on AI and enterprise security posture.
Staff-level engineer building security products for safe AI application and agent adoption at scale. Responsibilities include architecting secure systems across multiple teams, defining technical direction for AI security product areas, establishing security standards for reliability and maintainability, and identifying architectural risks. Requires 5+ years software engineering with technical leadership, deep proficiency in Go/Rust/Python, systems and security engineering expertise, and hands-on architecture of platforms spanning multiple teams. Preferred experience includes enterprise security products and AI/ML systems.
Southlake, Texas; Austin, Texas; Phoenix, Arizona, United States
Engineer responsible for securing AI systems across the enterprise. Develops and maintains internal AI security libraries, builds validators for prompt inputs and output guards against insecure execution, implements PII-protection and redaction features, and writes unit tests. Integrates security libraries with developer workflows, CI/CD pipelines, and AI/ML frameworks. Requires programming in Java, .NET, or Python; understanding of secure coding principles and OWASP; familiarity with GitHub and CI/CD tools; and exposure to cloud AI/ML frameworks (AWS Bedrock, Azure OpenAI, GCP Vertex AI).
Research position at Carnegie Mellon University's CERT Division focused on AI security. Responsibilities include vulnerability discovery and assessments for AI systems, reverse engineering AI models, evaluating effectiveness of AI system defenses, and identifying emerging threats. Work involves developing analysis approaches for AI system robustness, studying the AI security ecosystem, and publishing research findings for government and industry sponsors.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Conduct AI security research at Carnegie Mellon's CERT/SEI, focusing on vulnerability discovery, AI system robustness analysis, and defense evaluation. Reverse engineer AI systems, develop analysis methodologies and tools, study AI security ecosystems, and publish findings to influence national AI security strategy. Collaborate with cybersecurity experts, government officials, and industry practitioners on threat modeling, assessment frameworks, and emerging AI security challenges.
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Conducts research into vulnerabilities and defenses of AI/ML algorithms within the Secure AI Lab. Responsibilities include counter AI research to identify threat models and weak points in AI systems, designing novel attacks and defenses, and advancing adversarial machine learning. Leads research projects, develops prototype capabilities for government customers, mentors junior researchers, and publishes original research on AI algorithm vulnerabilities and security mitigations.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Establish and maintain application and AI security governance across enterprise systems. Conduct threat modeling for AI use cases and workflows, assess AI-related cybersecurity and data protection risks (prompt injection, model misuse, data leakage), review AI solution architectures for security design weaknesses, embed security requirements across AI development lifecycle, manage vulnerability assessment findings (SAST, DAST, API testing), and provide expert advisory on secure AI adoption. Requires 4+ years relevant experience, threat modeling expertise, and exposure to AI/LLM technologies through implementation, governance, or security assessments.
Principal Engineer responsible for technical strategy, architecture, and hands-on delivery of an enterprise AI Security Governance Program. Owns five control pillars: AI visibility & inventory, prompt & response protection, agent architecture & control, shadow AI & access control, and compliance & audit. Personally builds and matures security controls for AI-connected assets across enterprise-wide AI traffic and ServiceNow platform AI. Drives shift from manual security operations to AI-automated workflows, implements detection and response systems, data security posture management, and data-path controls. Provides technical leadership on AI data protection, agent permissions, and shadow AI risks to CISO and Audit Committee.
Remote role at Mercor focused on red-teaming conversational AI models. Responsibilities include diagnosing vulnerabilities, annotating model failures, and generating actionable adversarial data to improve AI system safety. Requires hands-on adversarial testing experience and technical understanding of AI safety principles.
Build and maintain the harness and platform infrastructure for NVIDIA's AI Safety & Security Engineering team that develops AI-powered tooling to find, validate, and patch software vulnerabilities. Design agent harness systems, evaluation infrastructure for running security research experiments, and reproducibility tooling. Partner with security researchers and evaluation engineers to support AI-driven vulnerability detection and validation workflows. Requires 5+ years software engineering experience, strong Python skills, and familiarity with agent frameworks, LLM orchestration, or ML infrastructure.
Capabilities researcher building Claude Security, responsible for identifying which security capabilities in frontier models are production-ready, measuring their performance on realistic cybersecurity tasks, and designing scaffolding to make them usable by non-expert security teams. Work includes rapid prototyping, rigorous evaluations, dataset and harness development for cybersecurity domains, and collaboration with engineers to operationalize capabilities. Requires 7+ years security research or engineering expertise in areas like vulnerability research, exploit development, malware analysis, or incident response, plus AI evaluation and red teaming experience.
Experienced AI Security Engineer securing AI and GenAI solutions across an international organization. Key responsibilities include implementing AI-specific security controls, protecting against prompt injection and adversarial attacks, designing LLM security guardrails, conducting AI security assessments, threat modeling, and establishing secure AI development and deployment practices. Requires 3-5 years security engineering experience and 2+ years hands-on AI/ML security background with cloud environment knowledge.
Application Security Engineer II at Abnormal AI, securing LLM-integrated features, agentic workflows, and model supply chains on AWS. Leads threat modeling and security architecture reviews for AI-powered systems, architects secure CI/CD tooling, designs automated security testing, responds to incidents, and coaches developers on secure coding for AI-native systems. Requires 5+ years of application security with expertise in prompt injection, model supply chain, and agentic-workflow risks, strong programming skills (Python/Go/Java/TypeScript), and web application security expertise.
Build lightweight security sensors and enforcement mechanisms for a GenAI security platform protecting against prompt injection, data leakage, and shadow AI attacks. Design adaptive agentic AI policy logic, develop endpoint and browser security components for real-time threat detection, optimize sensors for minimal system footprint, and own customer-facing technical problems end-to-end. Collaborate across security engineering, product, and customer teams to deliver real-time GenAI threat detection and data protection.
Expression of interest for technical roles focused on building agentic controls and forensics for future AI systems. Seeks AI research scientists and engineers with hands-on AI research experience to develop scalable approaches for understanding and mitigating misbehaviour in highly capable AI agents. Part of the UK AI Security Institute's mission to translate advanced AI risk knowledge into actionable government and developer guidance.
Lead AI security, governance, and risk management at BlinkRx. Design multi-layer AI security framework spanning data classification, detection/response, and agentic identity management. Conduct red-team assessments and adversarial testing of AI applications and workflows. Architect secure MCP server deployments and agent authentication standards. Define PHI/PII controls for AI systems. Integrate AI security controls into CI/CD pipelines. Develop automated testing for AI applications. Partner with Cloud Security and Engineering teams on AI workload guardrails in AWS.
Design and build low-level security sensors and enforcement engines for a Generative AI Security platform running on endpoints and browsers. Architect endpoint security solutions for macOS and Windows with OS-level APIs for real-time threat detection. Own the technical evolution of AI Security Posture Management engines, including adaptive policy logic and decision engines. Partner with security researchers to translate emerging AI threats into shipped protections. Requires 6+ years systems-level programming experience and deep understanding of OS internals, endpoint security frameworks, and browser extension development.
Develop state-of-the-art approaches for analyzing robustness and vulnerabilities in AI systems. Apply cybersecurity knowledge to understand how attackers exploit AI system weaknesses. Reverse engineer malicious code, design analysis methods and tools, identify emerging threats to AI systems. Study AI security ecosystems, evaluate effectiveness of defenses, develop threat models and datasets characterizing AI vulnerabilities. Publish research and support government-funded AI security initiatives.
AI Red Teaming
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Senior-level AI security engineer responsible for designing and implementing security controls for autonomous and multi-agent AI systems. Secures RAG pipelines, vector databases, and enterprise knowledge repositories. Performs threat modeling for AI applications and agent architectures, assessing risks from prompt injection, jailbreaks, data poisoning, and adversarial inputs. Implements guardrails, policy enforcement, and governance mechanisms for responsible AI. Designs monitoring and observability capabilities to detect abnormal agent behavior and model misuse. Partners with engineering teams to integrate AI security into development workflows and production deployments in regulated healthcare and federal environments.
Senior AI researcher building production-grade AI systems for vulnerability and exposure management at a cybersecurity startup. Responsibilities include designing AI-driven capabilities to analyze large-scale security data (assets, vulnerabilities, exposure context), owning the end-to-end AI lifecycle from research through production deployment, and developing LLM-based systems for security problem-solving. Requires 5+ years of applied AI/ML experience, hands-on LLM production expertise, and cybersecurity domain knowledge spanning asset management, vulnerability prioritization, and exposure analysis.
Build secure, reproducible infrastructure and tooling for AI security evaluations at 10a Labs. Design cloud-based environments, automation, and tooling supporting scalable adversarial testing of AI systems. Develop and automate attack simulations to validate security controls and improve resilience of AI-powered systems. Conduct threat modeling and security design reviews. Collaborate with red teamers to reproduce and remediate vulnerabilities uncovered during AI security assessments. Requires 3–5 years security/software engineering experience, hands-on offensive security work, Python, AWS/GCP, infrastructure-as-code, and familiarity with LLMs and AI security.
Own the security of AI adoption across the enterprise. Review and assess AI system integrations (LLMs, RAG, APIs, automation tools), conduct threat modelling, and develop AI-specific guardrails. Identify and mitigate prompt injection, model poisoning, data leakage, and adversarial attacks. Design data protection controls for AI contexts and DLP policies. Lead shadow AI governance, third-party AI tool due diligence, and compliance mapping against EU AI Act, NIST AI RMF, ISO/IEC 42001, and financial regulations. Author AI security policies, maintain AI risk registers, and report to management and auditors.
Design and build benchmarking and evaluation infrastructure for AI-powered security tooling at NVIDIA's AI Safety & Security Engineering team. Responsibilities include creating metrics and protocols to measure AI system effectiveness in vulnerability detection, designing reproducible experiments, mapping results to code and runs, and automating measurement infrastructure. Work directly with security researchers to evaluate agent and LLM behavior in security contexts.
Senior security engineer owning hands-on application security across product stack, including secure SDLC, CI/CD security tooling (Semgrep, CodeQL), vulnerability triage and remediation. Role includes growing responsibility for securing AI/LLM product surfaces and surfaces. Requires 3+ years hands-on security engineering experience with demonstrated build portfolio and ability to own scope autonomously.
Lead enterprise security strategy and architecture for AI-powered systems and Azure cloud environments. Establish security controls and runtime protections for agentic AI workflows. Conduct threat modeling for AI systems and cloud workloads, identifying risks from adversarial inputs and autonomous agents. Evaluate AI-generated code for security and define secure code review standards. Architect monitoring and hardening capabilities for AI agents at scale using Microsoft Defender and Sentinel. Embed security governance into AI/ML development pipelines and DevSecOps workflows. Research emerging AI and cloud security threats and translate insights into enterprise security improvements. Mentor teams on secure AI adoption and cloud security best practices.
Lead original technical research in AI safety, focusing on system understanding, control and alignment, and societal safeguards for large language models and agentic systems. Conduct interpretability research, steering vector development, and evaluation science. Balance peer-reviewed publication output with practical tooling and mitigation strategies. Collaborate with evaluation and red-teaming teams to inform frontier labs and government institutes on safe AI deployment.
AI Security Researcher at Carnegie Mellon University's Software Engineering Institute (CERT Division). Develops approaches for analyzing robustness and vulnerabilities in AI systems. Responsibilities include reverse engineering AI models and malicious code, vulnerability discovery and assessment, designing analysis tools and methodologies, evaluating effectiveness of AI security defenses, and conducting research on AI security threats and mitigations. Requires BS with 8 years, MS with 5 years, or PhD with 2 years in ML, cybersecurity, or related discipline. Requires practical vulnerability research experience, familiarity with AI/ML frameworks (PyTorch, TensorFlow, NumPy), reverse engineering tools, and programming proficiency.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute)AI lab
CERT Division researcher at Carnegie Mellon University focused on AI system security. Responsibilities include vulnerability discovery and assessment in AI systems, evaluating robustness of AI defenses, reverse engineering AI models, developing state-of-art analysis approaches, and shaping AI security best practices. Works with government sponsors, industry partners, and academic researchers. Requires 5+ years experience with PhDs, 8+ years with MS, or 10+ years with BS in ML/cybersecurity fields, plus proven expertise in both cybersecurity vulnerability research and practical AI/ML implementation.
Senior researcher at Carnegie Mellon's CERT Division focused on AI security. Responsibilities include developing approaches for analyzing robustness of AI systems, discovering vulnerabilities, reverse engineering AI models, evaluating defense effectiveness, and developing analysis tools and methodologies. Works on vulnerability assessments, threat characterization, and security research with government sponsors. Requires 10+ years experience (BS) or 8+ (MS) or 5+ (PhD) in ML/cybersecurity with practical vulnerability research, AI/ML implementation experience, familiarity with ML tools (PyTorch, TensorFlow, ART) and reverse engineering tools, and strong technical communication skills.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Lead a high-performing security engineering team at Isomorphic Labs, a frontier AI drug-discovery company. Own the technical security roadmap for AI-driven digital biology platforms, including ML pipeline security, model protection, autonomous workflow guardrails, and cloud infrastructure security. Translate complex scientific and business objectives into enterprise-grade security architecture while enabling velocity for ML researchers. Partner with the CISO on strategic initiatives spanning IAM, cloud security, secure SDLC, and security monitoring. Oversee incident escalation and vulnerability remediation for AI systems.
Lead technical incident response and security investigations across Twilio's global infrastructure, services, and applications. Responsibilities include triaging security events, containing and remediating incidents, automating response processes, and driving post-incident improvements. The role explicitly requires using AI and LLM-based tools to interpret alerts, reduce false positives, and manage AI model security threats including prompt injections, model evasion, and data poisoning. Conduct forensic analysis, maintain SIEM/SOAR platforms, and participate in on-call rotations.
Design adversarial prompts and conduct vulnerability assessment of frontier AI systems. Identify jailbreaks, unsafe behaviors, hallucinations, and policy failures across misinformation, cybersecurity, biosecurity, fraud, and political content domains. Evaluate model robustness, document vulnerabilities, contribute to safety benchmarking reports, and collaborate with AI researchers on model alignment and safety improvements. Requires 5+ years in AI Safety, Red Teaming, Trust & Safety, or cybersecurity.
AlphaSense seeks an AI Security Analyst to detect, investigate, and contain AI-related risks across enterprise AI tools, agents, and models. Responsibilities include monitoring AI usage end-to-end, hunting shadow AI and rogue agents, analyzing AI data exfiltration via DLP/CASB telemetry, monitoring developer AI tool governance, tuning detection rules, and supporting ISO 42001 compliance and audit evidence collection. The role requires 3–5+ years in security analysis or SOC work, hands-on experience in AI/LLM risk testing, CASB-based shadow IT/AI discovery, or agentic AI architecture security, and working knowledge of SIEM, DLP, threat detection frameworks, and AI-specific risk concepts.
Remote-Friendly (Travel Required) | San Francisco, CA
Red Team Engineer at Anthropic's Safeguards team conducting comprehensive adversarial testing of AI systems and products. Responsibilities include designing creative attack scenarios combining multiple exploitation techniques, researching novel testing approaches for agent systems and tool use, executing full kill-chain attacks simulating real-world threat actors, developing automated testing frameworks for continuous assessment, and translating security findings into product improvements. Requires penetration testing, red teaming, application security, model jailbreaking expertise, and hands-on experience with security testing tools and LLM-specific testing frameworks.
Security Engineer on Twilio's Security Incident Response Team (SIRT) responsible for leading technical response to security events across global infrastructure, services, and applications. Responsibilities include incident triage, containment, remediation, automation of manual processes, and developing scalable security response solutions. Desired qualifications include AI Model Security & Posture Management to prevent vulnerabilities like prompt injections and model evasion, and AI-Driven Threat Response using GenAI/LLMs for alert interpretation and threat contextualization. Requires 3+ years production incident response experience, SIEM/SOAR expertise, and cloud platform knowledge.
Remote-Friendly (Travel-Required) | San Francisco, CA | Seattle, WA | New York City, NY
Staff-level application security engineer leading M&A security due diligence and post-acquisition integration at Anthropic. Responsibilities include pre-close threat modeling and risk assessment, post-close SAST/DAST implementation, vulnerability remediation tracking, and coordinating cross-functional security integration. Role combines hands-on AppSec work (code review, threat modeling, penetration testing coordination) with AI-specific security focus, including securing agentic systems and LLM-integrated product surfaces. Develops automation tooling using LLMs to scale due diligence processes. Active participant in on-run incident response rotation and core AppSec projects.
Security engineering role focused on securing frontier AI evaluation infrastructure, including agent isolation, sandboxing, cloud platform hardening, identity and access management, and securing agentic systems. Responsibilities include designing and building isolation boundaries for untrusted AI-generated code, vulnerability remediation across AWS and Kubernetes environments, implementing least-privilege access controls for both people and agents, and code review. Requires 7+ years security engineering experience, deep AWS/Kubernetes knowledge, and strong systems/network/cloud security fundamentals.
Senior Security Analyst conducting hands-on security research on AI Agents and Non-Human Identity (NHI) systems in cloud and SaaS environments. Responsibilities include deep-dive vulnerability analysis of customer infrastructure using platform tools and manual techniques, identifying security gaps in AI agent permissions and machine identities, and delivering actionable remediation reports. Role bridges technical security research with product and sales impact, staying current on emerging AI agent and cloud security threats.
Senior AI Security Engineer at Isomorphic Labs focusing on securing frontier AI-powered drug discovery systems. Responsibilities include adversarial threat modeling tailored to AI/ML vulnerabilities, establishing protection frameworks for ML model artifacts, designing guardrails and sandboxing for LLM ecosystems and autonomous agentic workflows, collaborating on ML lifecycle security controls from training through inference, leading AI security incident response with machine learning-driven threat hunting, and automating compliance monitoring against emerging AI regulations. Requires deep understanding of ML frameworks, cloud infrastructure security, adversarial ML threat vectors, agentic framework security, production-grade security tooling development, and pragmatic risk assessment balancing rigorous security with research velocity.
Staff-level AI engineer responsible for designing, developing, and maintaining scalable GenAI-powered security capabilities within SentinelOne's unified security platform. Builds and integrates large language model and generative AI features into the security engine for real-time threat detection and response against AI-related threats. Leads end-to-end projects from requirements through production deployment, collaborating with data scientists to implement data integration processes and harden AI models for security operations. Requires 7+ years software development experience, strong Python proficiency, hands-on GenAI/LLM prototyping and production deployment, and a builder mentality for rapid iteration in a dynamic environment.
Detection & Response
LLM Security
ML Engineering
Prompt Security (now part of SentinelOne)Security vendor
Staff-level individual contributor driving technical strategy for application security across global product landscape. Responsibilities include threat modeling, vulnerability management at scale, DevSecOps automation, and pioneering AI-driven security automation using LLMs for code analysis and triage. Requires deep application security expertise, hands-on coding proficiency in modern languages, working knowledge of AI security frameworks, and ability to implement automated security guardrails that scale engineering velocity while maintaining security posture.
Develop machine learning-based prototypes, tools, and systems for AI security applications at Carnegie Mellon University's CERT Division. Collaborate with researchers to design and execute experimental AI security solutions, including AI red teaming and adversarial ML initiatives. Support processing and analysis of large cybersecurity datasets, apply software engineering best practices to build scalable systems, and translate security research concepts into practical operational capabilities. Work alongside elite AI and cybersecurity professionals on national security-relevant AI security challenges.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Senior ML researcher specializing in AI/ML algorithm vulnerabilities and defenses. Leads research in counter-AI threat modeling, adversarial machine learning, and defensive techniques. Develops and transitions security capabilities for government sponsors. Conducts original research in algorithm vulnerabilities, identifies attack vectors, designs novel mitigations, and evaluates defensive effectiveness. Mentors team members and collaborates on strategy for AI security research agenda.
Design and implement safeguards for internal AI usage, focusing on agentic systems, developer protection, and runtime security. Architect guardrails for tool-using AI systems including tool access controls, context isolation, and step-level validation. Build runtime enforcement mechanisms (interceptors, proxies, middleware) governing AI behavior at execution time. Design identity and access controls for non-human agents with short-lived credentials and scoped permissions. Implement logging and detection for AI activity including behavioral baselining and anomaly detection. Perform threat modeling using MAESTRO to identify attack paths and translate findings into practical safeguards. Protect developers using AI tools by preventing sensitive data exposure and validating AI-generated code.
Staff-level role designing and building autonomous AI agents and multi-agent frameworks for enterprise security operations at ServiceNow's Moveworks team. Responsibilities include architecting advanced agentic networks for proactive threat hunting, designing end-to-end incident response lifecycle automation, building MCP server orchestration systems, conducting purple team operations with internal AI red team agents, and validating automated detection and response pipelines. Acts as technical escalation point for complex incidents and strategic partner across security and infrastructure teams.
San Francisco, CA | New York City, NY | Seattle, WA
Lead Anthropic's security risk management as an engineering function. Build AI-native risk platforms using Claude to classify and triage risks. Drive end-to-end ownership of complex security problems across identity, secrets management, infrastructure, and secure frameworks. Partner with Security Engineering on threat modeling, risk quantification using methods like Monte Carlo simulation, and remediation roadmaps. Create dashboards and data pipelines for distributed risk ownership. Requires 8+ years security/software engineering experience, Python or systems language proficiency, and calibrated risk judgment.
Red Teaming Fellow conducting adversarial testing of frontier AI systems and models. Responsibilities include designing and executing red teaming exercises to identify vulnerabilities and failure modes, developing attack strategies and test cases, analyzing model behavior under challenging conditions, documenting findings, supporting evaluation framework development, and researching emerging AI threats and attack techniques.
Design and build ML systems for AI safety and model evaluation at 10a Labs. Develop evaluation pipelines, benchmarks, and metrics for frontier AI systems. Run ML experiments to assess capabilities, robustness, and limitations of advanced AI systems including LLMs and multimodal models. Train, fine-tune, and evaluate models; identify failure modes and security risks. Work with reinforcement learning, NLP, computer vision, and agentic AI fundamentals. Requires 3–5+ years ML experience, strong Python and PyTorch/JAX skills, experimental design expertise, and familiarity with adversarial evaluations and AI agent security risks.
Design and execute red team assessments against AI systems and applications. Develop adversarial testing methodologies to evaluate system security and robustness. Craft prompts, attack chains, and adversarial inputs to uncover unsafe behavior. Identify vulnerabilities and failure modes through offensive testing. Analyze system outputs to assess security and safety risks. Collaborate with engineers and researchers to improve system security. Requires 2–5 years of cybersecurity experience, proficiency with security tools (Splunk, Burp Suite, Metasploit, etc.), scripting capability, and familiarity with AI/ML attack vectors including prompt injection and jailbreak techniques.
Tel Aviv-Yafo, Ra'anana, Be'er Sheva, Yokne'am Illit, Israel
NVIDIA Networking's AI red teaming role focused on improving safety and security of AI models, systems, and infrastructure. Responsibilities include hands-on safety/security research, tool development to expose AI vulnerabilities, defining secure AI development processes, and partnering with cross-functional teams. Requires 5+ years in AI safety/security and offensive cybersecurity, Python expertise, and deep understanding of LLM, MLLM, generative AI, agents, and RAG workflows.
Senior security engineer role focused on securing AI-powered services. Responsibilities include designing and maintaining secure CI/CD pipelines with automated security controls, establishing secure software development lifecycle practices, conducting threat modeling and penetration testing, and assessing emerging AI-related risks including LLM applications, AI agents, prompt injection, and model supply chain security. Partner with engineering teams to evaluate and govern AI-powered development tools securely. Requires 4+ years cybersecurity/DevSecOps experience, hands-on software engineering knowledge, and practical experience with AI automation tools and agents.
Develop machine learning-based prototypes, tools, and systems for AI security applications. Collaborate with researchers to design and execute experimental AI security solutions. Support AI red teaming and adversarial machine learning initiatives. Process and analyze cybersecurity datasets including malware and incident data. Apply software engineering best practices to build scalable AI security systems. Requires 8+ years CS/ML/cybersecurity experience (or equivalent via MS/PhD).
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Senior AI Security Software Engineer at Carnegie Mellon University's CERT Division. Develops machine learning-based prototypes, tools, and systems for AI security applications. Supports AI red teaming and adversarial machine learning initiatives. Collaborates with researchers to design and execute experimental AI security solutions. Processes and analyzes cybersecurity datasets including malware and incident data. Requires 10+ years experience (CS/ML/cybersecurity background) or equivalent education. Work involves translating research concepts into operational capabilities impacting national AI and cyber security strategy.
AI Red Teaming
MLSecOps
Security Research
Carnegie Mellon University (Software Engineering Institute / CERT)AI lab
Develop machine learning-based prototypes, tools, and systems for AI security applications. Support AI red teaming, adversarial ML initiatives, and security analysis of AI systems including generative AI and LLMs. Process and analyze diverse cybersecurity datasets (malware, NetFlow, incident data) using ML and data analytics. Collaborate with elite AI and cybersecurity professionals to design and execute experimental AI security solutions and translate research concepts into practical operational capabilities.
Director leading LawZero's Evaluations Team to assess safety and capabilities of AI systems (Scientist AI and frontier LLMs). Responsibilities include designing evaluation strategies, building datasets and benchmarks for safety properties, overseeing red-teaming programs (automated and manual), evaluating adversarial robustness (jailbreaks, prompt injection, data poisoning), constructing evaluation infrastructure at scale, and communicating results to AI safety community. Requires 10+ years ML experience with 5+ years leading technical teams, hands-on expertise in LLM/ML safety evaluations, and track record building evaluation datasets and interactive environments for safety-relevant properties.
Senior-level role at CERT/SEI developing machine learning-based prototypes, tools, and systems for AI security applications. Responsibilities include AI red teaming, adversarial machine learning research, security analysis of AI systems, and processing cybersecurity datasets. Work spans generative AI, large language models, and counter-AI applications. Requires BS+10yr, MS+8yr, or PhD+5yr in CS, ML, cybersecurity, or related fields. Demands security clearance and software engineering expertise with containerization and microservices.
Deep technical security researcher focused on discovering novel risks and attack vectors in modern cloud and AI-native architectures. Conducts independent research projects to identify unaddressed risk areas, delivers proofs of risk and technical POCs, and collaborates with Product and Engineering teams to translate research into product capabilities. Requires 5+ years hands-on security research experience, proficiency in Python/Go scripting and query languages (KQL, SQL), and specialized knowledge in AI security risks as deployed in enterprise environments.
Conduct deep technical research to discover novel risks and attack vectors in cloud- and AI-native environments. Work with Product and Engineering teams to identify unaddressed risks, deliver proofs of concept, and define foundational security capabilities. Requires 5+ years hands-on security research experience, proficiency in Python/Go scripting, and preferred expertise in AI security risks in enterprise deployment contexts.
Conduct applied and theoretical research into AI vulnerabilities, attack/defense methods, and security/safety benchmarking for GenAI systems. Investigate adversarial attacks, data poisoning, prompt injection, jailbreaks, and multi-modal vulnerabilities. Design and evaluate mitigation strategies including fine-tuning, guardrails, and content filtering. Develop AI security evaluation frameworks and datasets. Integrate findings into Vulcan AI security products and publish in top-tier venues.
Lead ML team developing and operationalizing models for GenAI threat detection, prompt injection prevention, and vulnerability assessment. Responsibilities include fine-tuning LLMs for security intents, scaling MLOps pipelines for security production environments, monitoring model drift in security context, and bridging research findings on GenAI threats into robust detection systems supporting both Blue Team (guardrails) and Red Team (automated vulnerability assessment) capabilities.
Lead ML team responsible for operationalizing GenAI threat detection models that power security guardrails and automated vulnerability assessment. Develop and fine-tune LLMs to detect prompt injection and multimodal threats. Establish MLOps infrastructure with rigorous model monitoring and data governance for production security environments. Bridge research findings on GenAI threats into robust, deployable models. Mentor ML engineers and manage GPU resources for both training and inference workloads in security-critical applications.
Senior AI Security Engineer responsible for red teaming and penetration testing of generative AI implementations including LLMs, agents, chatbots, and coding assistants. Combines offensive testing (prompt injection, jailbreaking, data exfiltration, agent abuse) with defensive work (guardrails, input/output controls, system prompt hardening). Designs and operationalizes security controls based on OWASP Top 10 for LLM Applications and Agentic Applications. Develops automation tools for continuous LLM testing, documents findings with remediation playbooks, and collaborates with SOC on detection. Requires 4+ years in application security/penetration testing with hands-on LLM and generative AI attack/defense experience.
Conduct AI security research at Carnegie Mellon's Software Engineering Institute (CERT Division). Develop approaches for analyzing robustness and identifying vulnerabilities in AI systems. Perform reverse engineering of AI systems and malicious code, evaluate effectiveness of AI security defenses, and design analysis tools and methodologies. Study AI security ecosystems, threat characterization, and disclosure practices. Collaborate with government, industry, and academic partners on AI security strategy. Requires BS in ML/cybersecurity/related field with 3+ years experience OR MS with 1+ year experience. Must have practical vulnerability research background, AI/ML implementation experience, familiarity with ML frameworks (PyTorch, TensorFlow, ART) and reverse engineering tools, and strong technical communication skills.
London, UK (hybrid, with UK office locations); US remote possible
Research role focused on model transparency and oversight reliability at the UK AI Security Institute. Develop detection and mitigation methods for evaluation gaming, model deception, and misalignment in frontier AI systems. Conduct red-teaming, auditing, and interpretability research on large language models. Build research infrastructure, tooling, and agent orchestration systems. Collaborate with frontier AI labs (Anthropic, OpenAI, DeepMind) and UK government on safety case reviews and alignment evaluation methodology.
New York City, NY; San Francisco, CA; Washington, DC; Remote-Friendly
Hands-on Threat Intelligence Engineer responsible for detecting and countering nation-state and criminal threats targeting frontier AI development. Tracks adversaries, builds automated pipelines and tooling to operationalize indicators into detection and alerting stacks, conducts intelligence-driven threat hunts across endpoint/cloud/identity telemetry, performs technical malware and infrastructure analysis, authors detection rules, and partners with detection engineering and incident response teams. Requires 5+ years in cyber threat intelligence or intrusion analysis, deep knowledge of threat actors and tradecraft, strong Python engineering, malware analysis expertise, and detection logic authoring experience.
Hands-on security infrastructure engineer responsible for vulnerability management, SIEM/XDR operations, incident response, supply chain security, and detection engineering across cloud and application systems. Requires AWS and Kubernetes expertise with explicit mandate to leverage AI/LLM tools (Claude) for security automation, detection rule tuning, and operational efficiency. Performs root cause analysis, threat modeling, and drives CI/CD hardening and container security initiatives.
Engineering fellowship at AI safety and threat-intelligence company 10a Labs. Hands-on role combining applied research and engineering across red teaming, abuse detection, and AI system evaluation. Three concentrations available: Software Engineering (cloud infrastructure, testing, secure systems), Data Engineering (data sourcing, architecture, dashboards), or Machine Learning (model training, agentic systems, frontier model evaluation). Supports client-facing products and research initiatives in AI security and Trust & Safety.
London (remote-friendly with UK/EU timezone overlap required)
Member of Technical Staff in Safety for Agents team at Cohere, focused on developing safer, more secure LLMs that access external resources and take actions. Responsibilities include designing data generation and post-training algorithms, creating evaluation methods for agent safety, analyzing dataset quality and biases, hands-on LLM training on distributed infrastructure, and collaborating with ML and product teams to improve model robustness and generalizability. Requires strong software engineering, statistical expertise, and ML research background.
Research Scientist focused on agent robustness and AI safety. Responsibilities include researching AI agent capabilities with emphasis on safety and risk factors, designing harnesses to test agents' tendency to take harmful actions, creating exploits and mitigations for agent failure modes, and characterizing risks in multi-agent systems. Requires practical ML research experience, proficiency with post-training techniques (RLHF, DPO, GRPO), published research track record, and 3+ years addressing sophisticated ML problems. Red-teaming and adversarial testing experience valued.
Research Scientist role focused on AI controls and monitoring at Scale Labs. Responsibilities include designing real-time monitoring and observability methods for advanced AI models, researching layered control mechanisms including fail-safes and intervention protocols, conducting red-team simulations to identify gaps in oversight, and collaborating with policymakers and engineers on AI safety standards. Requires 3+ years ML experience, published research track record, and hands-on prototyping skills. Nice-to-haves include runtime monitoring, anomaly detection, alignment research, and post-training techniques like RLHF.
Lead Machine Learning team developing GenAI security models for threat detection and automated vulnerability assessment. Operationalize security research findings into production LLM-based detectors for prompt injection and GenAI threats. Fine-tune language models using LoRA/PEFT for multilingual security detection. Implement MLOps pipelines with data governance, model monitoring, and drift detection. Collaborate with security research and platform teams to integrate threat detection models into Vulcan's blue team (defensive) and red team (vulnerability assessment) capabilities. Prepare multimodal detection systems for visual/audio threats. Lead ML engineers and manage compute resources.
Applied data scientist at 10a Labs focused on designing and automating adversarial red-teaming frameworks and evaluations for frontier AI systems. Responsibilities include developing evaluation frameworks, designing red-teaming strategies, leading adversarial testing initiatives (jailbreak simulation, evasion probes), automating workflows for model evaluation and AI experiments, fine-tuning LLMs for safety assessment, and coordinating with ML and infrastructure teams. Requires 3-5 years production ML experience with emphasis on model behavior analysis and safety evaluation.
Design and build automated pipelines for pre-deployment training-run assessments of frontier AI models. Conduct red-teaming and evaluation of model checkpoints at various post-training stages to detect misalignment, scheming, and undesirable behaviors. Develop evaluation methodologies and infrastructure for automated analysis. Run bi-weekly pre-deployment evaluation campaigns with frontier labs, perform targeted follow-up experiments, and report findings. Requires strong Python software engineering, data analysis, anomaly detection, and AI power-user capabilities.
Research Scientist/Engineer at Apollo Research focused on developing a "Science of Scheming." Build evaluations and empirical foundations for detecting deceptive alignment and studying how scheming risks scale with model capability. Design novel evaluation techniques for frontier AI systems, train model organisms via RL, analyze AI reasoning processes, and collaborate with leading AI labs to inform safe deployment. Requires empirical research design, strong software engineering (Python), hands-on RL/LLM experience, and rigorous quantitative analysis.
San Francisco, CA | New York City, NY | Seattle, WA
Staff-level security engineer responsible for designing and building complex security systems protecting Anthropic's AI infrastructure. Focus areas include identity and secrets management for model weights and training data, developer security and supply chain hardening, infrastructure security across multi-cloud environments, and secure cryptographic frameworks. Leads threat modeling efforts, mentors security engineers, and drives security ownership across the engineering organization. Requires 8+ years of software engineering with deep security expertise, strong systems programming skills, and experience with identity systems, cryptography, Kubernetes security, and large-scale infrastructure.
Lead Machine Learning team at AIFT developing GenAI security detection models. Responsibilities include: operationalizing threat detection techniques from security research into production ML models; fine-tuning LLMs for multilingual security intents; evaluating multimodal (text, image, audio) threat detection; optimizing ML pipelines and infrastructure; implementing data versioning and monitoring for model drift; collaborating with research and platform teams on model integration; and translating technical constraints to leadership. Requires 5+ years ML engineering experience, MLOps proficiency (MLflow, Kubeflow, Airflow, DVC), Python/Docker/Kubernetes expertise, and LLM/Transformer knowledge.
LLM Security Evaluation Expert role at government solutions firm focused on rigorously testing security and integrity of Large Language Models. Responsibilities include designing sophisticated adversarial prompt attacks, testing model vulnerabilities, documenting security findings, and identifying model biases. Requires TS/SCI clearance and deep LLM knowledge combined with cybersecurity expertise. Emphasis on responsible adversarial testing practices to safeguard AI systems.
Senior Generative AI Analyst focused on leading advanced safety red-teaming efforts for generative AI systems including LLMs and text-to-image models. Responsibilities include designing and executing adversarial testing, developing adversarial prompts to identify model vulnerabilities, analyzing safety insights, and contributing to red-teaming tools and methodologies. Works across frontier model labs, enterprises, and UGC platforms to deliver rigorous testing strategies and vulnerability assessments.
Subject Matter Expert on AI security within Greenhouse's product security team. Responsibilities include securing emerging AI and Machine Learning features, improving security best practices across agile SDLC, collaborating with engineers to harden AI/ML systems, staying current on AI/ML security trends, and proactively identifying and mitigating risks before deployment.
Build foundational systems for safety and oversight of AI models at Anthropic's Safeguards team. Develop infrastructure for detecting unwanted model behaviors, preventing misuse, and supporting human/agentic review. Create robust multi-layered defenses for real-time safety improvements and establish metrics and evaluation systems for AI system safeguards.
Design, build, and evaluate technical safeguards for AI systems to mitigate chemical-biological misuse risks. Lead research projects assessing the effectiveness of mitigations like biological/chemical classifiers and data filtering. Partner with frontier AI developers to stress-test safeguards, critique capability assessments, and translate findings into actionable guidance for improving developer practices and government policy.
AI Red Teaming
Security Research
UK AI Security Institute (UK Government)Government
Remote-Friendly (Travel-Required) | San Francisco, CA | Washington, DC
Detect and investigate misuse of Anthropic's AI systems for cyber operations including malware development, social engineering, and influence operations. Develop abuse detection signals, create threat intelligence reports on LLM attack vectors, conduct cross-platform threat analysis, and implement systematic safety improvements. Build threat detection systems and maintain relationships with external threat intelligence partners.
Lead and scale multi-disciplinary AI security teams focused on red teaming, adversarial testing, and guardrail protections for GenAI and agentic AI systems. Oversee advanced adversarial evaluations, define AI red-teaming frameworks aligned with OWASP AI and NIST guidelines, operationalize automated red-team engines, and partner with product and engineering to design enterprise-ready AI guardrails with policy enforcement, monitoring, and anomaly detection. Requires 12+ years in cybersecurity/ML security and extensive leadership experience building high-performance teams.
Build and ship cybersecurity products powered by Claude AI models. Develop agent loops, tool integrations, and infrastructure for security applications. Collaborate with research teams to identify model capabilities for security use cases and iterate with customers on security solutions across the full stack from prototype to production.
Build intelligent generative AI agents and ML-based detection models for cybersecurity threat analysis and automated investigation workflows. Design feature representations, file parsers, and malware detection engines in C++ and Python. Combine deep security research with modern AI techniques to develop production-ready, customer-facing security capabilities. Serve as cybersecurity expert within data science team, guiding threat modeling and security-driven AI design. Requires 5+ years cybersecurity research, malware analysis, EDR, and threat detection experience, plus strong ML engineering skills.
Research Engineer at Anthropic's Horizons team, specifically the Cybersecurity RL team. Develops reinforcement learning systems and novel approaches to advance AI model capabilities in secure coding, vulnerability remediation, and defensive cybersecurity. Responsibilities include designing and implementing RL environments, conducting experiments and evaluations, delivering work into production training runs, and collaborating with researchers, engineers, and cybersecurity specialists. Requires domain expertise in cybersecurity (security research, fuzzing, detection and response, or applied defensive work) combined with machine learning experience and strong software engineering skills.
Security researcher responsible for proactively red teaming and hacking high-impact GenAI systems at Microsoft, including Bing Copilot, Security Copilot, Github Copilot, and Windows Copilot. Conducts adversarial machine learning research to find safety and security failures in AI models and applications pre-launch. Works with interdisciplinary teams of red teamers, ML researchers, and safety experts to identify and inform mitigations for AI security and trust risks.
Evaluate and red team large language models and AI agents for safety issues and security vulnerabilities. Create repeatable, offline test cases with automatic evaluation. Build automation scripts, custom tools, test environments, and harnesses. Design realistic attack scenarios, document findings with reproduction steps, and work with LLM-specific risks including prompt injection and OWASP Top 10 for LLMs. Contribute to security research and advise on secure coding and platform hardening.
Lead end-to-end AI red teaming security assessments for enterprise clients, focusing on text-based LLM applications and multimodal agentic systems. Develop and improve red teaming methodologies and automated tooling for security testing. Conduct comprehensive assessments covering LLM vulnerabilities including prompt injection, data poisoning, and jailbreaking. Author detailed security reports and present findings to stakeholders. Stay current with emerging AI attack vectors and translate cutting-edge research into practical testing approaches.
San Francisco, CA; New York City, NY; Seattle, WA; Washington, DC
Detection and response engineer at Anthropic responsible for building solutions to monitor for threats, investigate incidents, and coordinate response efforts. Develops and deploys novel tooling that may leverage Large Language Models to enhance detection, investigation, and response capabilities. Creates detections, playbooks, and workflows to identify and respond to incidents. Leads cybersecurity incident response efforts across diverse domains from external attacks to insider threats. Requires 3+ years software engineering or 5+ years detection/incident response experience, cloud environment knowledge, Python proficiency, and security operations experience.
Senior security consultant role focused on AI/ML security architecture and design. Responsibilities include defining enterprise AI/ML security strategy, designing secure AI/ML models and pipelines across cloud platforms, leading adversarial testing and risk assessments, and embedding ethical AI governance. Works with major government and enterprise clients on securing advanced AI/ML systems against adversarial threats and compliance challenges.
Research Scientist role at UK AI Security Institute focused on AI security for chemical and biological risks. Lead technical work on capability thresholds, ecosystem engagement, and mitigation strategies for frontier AI models including LLMs. Responsibilities include evaluating frontier AI developer safeguards, collaborating on risk mitigation projects (e.g., biological/chemical classifier robustness evaluations), developing evaluation methodologies, defining risk/capability thresholds, and engaging with academic and industry stakeholders on AI safety implications. Requires background in biological/chemical science or ML, plus experience in biosecurity, AI governance, or frontier AI risk mitigation strategies.
LLM Security
Security Research
UK AI Security Institute (UK Government)Government
Security researcher focused on discovering and responsibly disclosing vulnerabilities in GenAI applications, LLMs, RAG pipelines, and multi-agent systems. Responsibilities include designing novel attacks, creating proof-of-concept exploits, authoring peer-reviewed papers for top-tier conferences (Black Hat, DEF CON, RSA), and leading public workshops on AI security techniques. Requires 4+ years of offensive security or adversarial ML experience, demonstrated jailbreaking capabilities on commercial models, strong Python skills, and proven publication track record.
Software engineer on Anthropic's Safeguards team building safety and oversight mechanisms for AI systems. Develops monitoring systems to detect unwanted model behaviors, builds abuse detection infrastructure, and creates multi-layered defenses for real-time safety improvement at scale. Surfaces abuse patterns to research teams for model hardening and enforces terms of service and acceptable use policies. Requires 8+ years software engineering experience and expertise in detection, adversarial inputs, and AI/ML trust and safety systems.
Build systems to detect and mitigate misuse of AI systems, including developing classifiers for harmful use detection, monitoring coordinated attacks, evaluating safety of agentic products against prompt injection and other threats, and conducting automated red-teaming research. Responsibilities span synthetic data pipelines, anomaly detection at scale, multi-exchange harm monitoring, threat modeling for agentic risks, and adversarial robustness research. Requires 4+ years ML/research engineering experience and ability to work across research-to-deployment pipeline.
Design and execute adversarial security tests against large language models, AI agents, and RAG pipelines. Create and automate attack prompts including prompt injection and jailbreak attempts. Build test suites, define scoring rubrics for model responses to adversarial inputs, and document reproducible findings with risk ratings and mitigations. Assess attack surfaces involving data exfiltration, web systems, APIs, and infrastructure. Contribute Python, Bash, or PowerShell scripts to scale testing operations.
Design and implement security frameworks for GitLab's internal AI systems, including LLMs and AI agents. Secure Model Context Protocol implementations, manage non-human identity systems with zero-trust access controls, govern internal AI tool usage through DLP and monitoring, and prevent data leakage through AI systems. Collaborate cross-functionally on AI security adoption and evaluation of enterprise AI tools.
Human data expert who probes AI models and agents with adversarial inputs to surface vulnerabilities, generate attack datasets, and create red-team data for AI safety. Responsibilities include testing for jailbreaks, prompt injections, misuse cases, and exploits; annotating failures; classifying vulnerabilities; and documenting reproducible attack cases and reports.
Design, build, and deploy AI-powered security agents for threat detection, incident analysis, and automated response. Develop agentic workflows that autonomously perform log forensics, threat hunting, and incident response tasks. Integrate agents with data lakes, SIEM platforms, and security toolstacks. Requires 5+ years professional experience with 3+ years in cybersecurity operations, hands-on incident response, Python proficiency, LLM/agentic framework experience, and cloud security knowledge.
Join the AI Safety Institute's Safeguard Analysis Team to research and develop interventions that secure frontier AI systems from abuse. Responsibilities include assessing security threats to advanced AI systems, developing novel attacks against LLMs, conducting red-teaming evaluations, and building infrastructure for running security assessments. The role spans research-oriented work (threat analysis, attack development) and engineering-oriented work (evaluation infrastructure). Candidates should have experience in AI security, computer security, machine learning, red-teaming, and proficiency with large language models and Python. Mentorship from world-class researchers including alumni from Anthropic, DeepMind, and OpenAI.
Threat Detection Researcher at Wiz responsible for developing cloud-native threat detection capabilities. Drives the full lifecycle of security detections from research and data analysis to production deployment. Designs behavioral baselines for cloud environments, expands detection engine with novel telemetry sources, and conducts deep technical research into cloud services to uncover attack vectors. Investigates real-world attacks and hunts emerging threats targeting cloud ecosystems. Requires 6+ years hands-on security/threat research experience, Python/Go proficiency, and proven track record of delivering actionable security impact.
Join Wiz's Threat Research team to develop cloud-native threat detection capabilities. Drive the complete lifecycle of security detections from research and data analysis to production deployment. Design behavioral baselines for complex cloud environments, develop high-fidelity detections from diverse signals, and expand the detection engine with novel telemetry sources. Conduct deep technical research into cloud services to uncover attack vectors, investigate real-world attacks across cloud platforms and identity providers, and hunt emerging threats targeting the cloud ecosystem. Requires 6+ years of hands-on security/threat research experience, proficiency in Python, Go, and query languages, plus background in incident response, red teaming, or threat hunting.
Detect and respond to malicious uses and abuse of AI systems. Investigate threat actors and harmful behaviors through SQL and Python data analysis, OSINT cross-referencing, and detection pipeline development. Support incident escalations and provide investigative reports for frontier AI labs and platforms. On-call responsibilities for urgent security escalations.
Research Engineer on Alignment Science at Anthropic focused on exploratory experimental research on AI safety. Responsibilities include building and running ML experiments to understand and steer AI system behavior, testing robustness of safety techniques through adversarial model training, running multi-agent RL experiments, building tooling to evaluate LLM-generated jailbreaks, and contributing to alignment stress-testing and pre-deployment safety assessments. Work feeds into key safety efforts including the Responsible Scaling Policy and involves collaboration with Interpretability, Fine-Tuning, and Red Team colleagues.
New York City, NY; Seattle, WA; San Francisco, CA; Washington, DC
Build and scale next-generation AI-powered security analytics infrastructure. Design and implement scalable data pipelines for processing security telemetry, develop ML-powered detection systems, and create solutions leveraging Claude for security operations. Architect storage and querying of large security datasets, develop detection capabilities from development to incident response, and mentor security engineers. Requires 7+ years software engineering experience with focus on security/infrastructure/data pipelines, strong background in detection engineering or security operations, and proven track record applying ML/AI to security problems.
Remote-friendly (San Francisco, Seattle, New York)
Lead application security for Claude and related AI systems at Anthropic. Design and operate LLM-driven security systems including automated code analysis, vulnerability remediation, and threat modeling. Secure novel AI infrastructure including multi-agent orchestration, sandboxed code execution, and delegated credentials. Own security systems end-to-end, conduct threat modeling for agentic systems, operate bug bounty programs with AI-assisted triage, and respond to incidents in production AI systems. Use Claude as a primary security tool while providing human judgment and system-level reasoning.
Application Security Engineer at xAI responsible for securing cloud-native applications and AI systems. Conduct code reviews, implement secure coding guidelines, integrate security in CI/CD pipelines, perform threat modeling, and manage vulnerability remediation. Address AI/ML-specific security concerns including OWASP LLM Top 10. Evaluate software supply chains and SBOMs. Support incident response and stay current on emerging threats in cloud-native and AI technologies.
Conduct in-depth research on AI-specific security threats including adversarial attacks, model tampering, and data privacy issues. Develop and implement strategies to detect and mitigate AI security vulnerabilities in NLP, computer vision, and other ML domains. Collaborate with cross-functional teams to integrate AI security measures into products and stay current with AI security research trends. Requires 5+ years AI/ML experience with demonstrated security research publications and hands-on experience deploying secure AI systems.
ByteDance seeks an AI Security Researcher to conduct research and analysis of security threats to AI technology. Responsibilities include identifying vulnerabilities in AI systems, developing countermeasures, conducting penetration testing and risk assessments, monitoring system security, developing security protocols, analyzing potential security incidents, and collaborating with cross-functional teams to implement security measures. The role requires expertise in AI security with focus on safeguarding cutting-edge AI systems and staying current with industry threats and compliance standards.
Security engineer specializing in red teaming and penetration testing for Cymetrics, a cybersecurity SaaS platform provider. Responsibilities include planning and executing red team exercises, conducting penetration tests across modern web frameworks (React, Angular, Vue.js), performing lateral movement and network reconnaissance, researching vulnerabilities in websites and open-source projects, assisting with automated security tool development, and collaborating on proprietary security product improvements. Requires 3+ years red teaming experience, expertise in OWASP methodologies, web vulnerability assessment, and client communication. The role supports a comprehensive security assessment SaaS platform that includes AI model vulnerability verification.
AI Red Teaming
Application Security
Security Research
aiFT (Vulcan)Security vendor
No roles match your filters
Try a broader search term, or clear the filters to see every open role.
Hiring for an AI security role?
Send us the posting and we will add it to the board
