Model Policy Manager

📋 Description

• Design and maintain model policies across safety domains.
• Translate risk models into behavioral specs, eval criteria, safeguards.
• Define boundaries between beneficial uses of AI and harmful outcomes.
• Build policy artifacts to support training, evaluation, deployment; collaborate with teams.
• Use red-teaming results, deployment data, failures, and edge cases to improve policy.
• Identify emerging frontier-risk areas where AI may create safety challenges.

🎯 Requirements

• Have strong judgment about AI risk in ambiguous, high-impact areas.
• Experience building or applying policies, taxonomies, harm/threat models.
• Can move across domains and know when to seek expert input.
• Turn fuzzy questions into structured policy frameworks and enforcement.
• Experience using empirical evidence (evaluations, red-teaming) to inform policy.
• Think in systems across policy, data, graders, classifiers, training, safeguards.

🎁 Benefits

• Hybrid work: 3 days in SF office, optional remote Thursdays and Fridays.
• Relocation support for new hires.
• Modern office amenities: meals, snacks, nap rooms, private outdoor space.
• Private bike storage and other on-site facilities.