[EJV] Penetration Tester — AI Safety & Security
Core
Penetration Tester executing safety and security evaluations on LLM deployments, agentic AI systems, and tool-calling integrations to ensure safe behavior and resistance to manipulation.
Role type
individual-contributor penetration tester (AI safety & security)
Builds
AI-enabled products and internal tooling (LLM deployments, agentic AI systems, MCP-style integrations)
Domain
Artificial Intelligence / Machine Learning Security
Deliverable
production ML models | product features
Required skills
penetration testing, LLM security testing, prompt injection testing, jailbreak testing, adversarial testing, Python scripting, JavaScript/Node scripting, technical writing, OWASP LLM/Agentic Top 10 knowledge
Preferred skills
Model Context Protocol (MCP) familiarity, traditional web/API/network pentesting, AI governance concepts (NIST AI RMF, ISO/IEC 42001, EU AI Act), open-source security tooling contributions
Technologies
Python, JavaScript, Node, MCP, OWASP LLM Top 10, OWASP Agentic Top 10, MITRE ATLAS
Responsibilities
Execute penetration tests and safety evaluations against LLM-based features and AI agents; Run red-teaming and jailbreak testing to surface harmful outputs; Test for prompt injection, excessive agency, and tool/plugin abuse; Verify human-in-the-loop and guardrail controls under adversarial conditions; Write clear, reproducible findings reports; Help build and maintain test tooling (adversarial-prompt scripts, fuzzing scripts); Track published AI security research and integrate new attack techniques; Work with AI/ML engineering teams to reproduce and confirm fixes; Support team lead in scoping external red-teaming engagements
Seniority
Mid-level, hands-on IC