GenAI Red Teamer

The work

Run structured adversarial testing for LLM applications — jailbreaks, policy bypass attempts, and tool-abuse scenarios aligned to OWASP LLM Top 10-style risks. Outcomes are findings and regression tests, not marketing claims about eliminating risk.

You will

• Design adversarial prompt and tool-use scenarios per application threat model
• Convert repeatable findings into automated guardrail and policy regression tests
• Coordinate with eval engineers so safety suites share goldens and CI infrastructure
• Report residual risk in plain language for non-security stakeholders

You likely have

• 4+ years in application security, ML safety, or adversarial testing
• Familiarity with LLM-specific failure modes (prompt injection, tool abuse, data leakage paths)
• Disciplined documentation — no sensationalism in client-facing artefacts
• Ability to work without overstating coverage or guarantees