💼 About This Role

You'll define model behavior policies for OpenAI's frontier AI systems in high-risk contexts like agentic and multimodal systems. Your work will directly shape safety guardrails that protect users while enabling beneficial AI use, translating ambiguity into measurable safety criteria.

🎯 What You'll Do

• Design and maintain model policies across safety-relevant domains.
• Translate risk models into behavioral specifications and evaluation criteria.
• Use red-teaming and deployment data to improve policy quality.
• Contribute to system cards, safety reports, and launch reviews.

📋 Requirements

• strong judgment about how advanced AI systems affect real-world risk
• experience building policy frameworks for complex technical systems
• ability to move across domains without deep expertise in every area
• comfort using empirical evidence to inform policy decisions

✨ Nice to Have

• Experience with red-teaming or threat modeling
• Background in AI safety research
• Knowledge of adversarial machine learning

🎁 Benefits & Perks

• 🏖️ Hybrid work model (3 days in office)
• 💰 Equity offered
• 🍽️ Three in-house prepared meals daily
• ☕ Well-stocked kitchens with snacks and drinks
• 🚲 Private bike storage

📨 Hiring Process

Estimated timeline: 3-5 weeks · AI estimate

1. 1Application Review· 1-2 weeks
2. 2Technical Interview· 1 hour
3. 3Onsite Interviews· 3-4 hours