Requirements

Ability to design and execute fast-paced empirical research experiments with rapid iteration.
Expertise as an AI power-user, proficient with multiple models and AI workflows.
Strong conceptual thinking and scientific writing skills, familiar with AI scheming literature.
Intense interest in AI progress and cognition, continuously updating knowledge on new models.
Hands-on experience training LLMs via reinforcement learning, troubleshooting technical issues.
Strong analytical skills in quantitative fields such as scaling laws, statistical physics, or applied statistics.
Commitment to the mission of reducing AGI misalignment risks with a dedicated work culture.

Nice to Haves

Strong software engineering skills, particularly in Python.
Project management experience, capable of keeping research projects on schedule.

Collaborate with leading AI developers across multiple labs to influence AI system development.
Study reinforcement learning dynamics related to reward-seeking, evaluation awareness, and misaligned preferences.
Investigate scaling laws of scheming and predict how scheming risks evolve with model capabilities.
Develop novel evaluation techniques capable of scaling to evaluation-aware AI models.
Discover new patterns in AI cognition and reasoning processes of frontier systems.

Market-competitive salary and equity with different ranges depending on role seniority.
Unlimited vacation and sick leave.
Up to 6 months paid parental leave.
Comprehensive health, dental, and vision insurance.
Retirement savings with competitive employer matching (e.g., 401(k) for US employees).
Provided meals on workdays including lunch, dinner, and snacks.
Paid work trips including staff retreats, business trips, and conferences.
Yearly $1,000 professional development budget.
Relocation support and visa sponsorship for UK and US roles.