Head of Evals, AI Red Teaming

Lead evaluation quality and design for AI red-teaming safety tests deployed to frontier model developers.
- Review and approve red-teaming tasks, transcripts, and grades to ensure evals teach models robust defenses against prompt injection.
- Design new evaluation methodologies and environments targeting gaps in model robustness, informed by failure analysis.
- Build agent-based automation tools and checkers to scale your review judgment without proportional headcount growth.
- Requires calibrated eval judgment, fluency with LLMs and coding agents, and sustained attention to detail across hundreds of reviews.
- Desirable: prior experience evaluating benchmarks, building LLM judges, designing eval environments, or prompt injection and red-teaming work.

Lead evaluation quality and design for AI red-teaming safety tests deployed to frontier model developers.
- Review and approve red-teaming tasks, transcripts, and grades to ensure evals teach models robust defenses against prompt injection.
- Design new evaluation methodologies and environments targeting gaps in model robustness, informed by failure analysis.
- Build agent-based automation tools and checkers to scale your review judgment without proportional headcount growth.
- Requires calibrated eval judgment, fluency with LLMs and coding agents, and sustained attention to detail across hundreds of reviews.
- Desirable: prior experience evaluating benchmarks, building LLM judges, designing eval environments, or prompt injection and red-teaming work.

Trajectory Labs is a startup that builds RL environments for frontier AI labs to train robust, secure, and reliable models.