Product Manager, Safety Research
San Francisco, CANew York, NYProduct ManagementRemote5+ YOE
Summary
Bridges AI safety research and North product, translating model evaluations and threats into safety features, roadmaps, and guardrails. Requires 5+ years PM/research ops experience with ML teams and technical depth in AI safety.
About the role
Responsibilities
- Serve as product bridge between safety research teams and North, translating findings into guardrails and safeguards.
- Own safety product roadmap, prioritizing based on research, misuse patterns, threats, and customer needs.
- Partner with modeling teams to scope and interpret safety evaluations across adversarial inputs and edge cases.
- Define evaluation frameworks to assess safety properties and detect regressions.
- Coordinate guardrails and interventions across research, engineering, and policy.
- Monitor AI safety landscape (prompt injection, jailbreaks, agentic misuse) and reflect in roadmap.
- Build processes for scaling safety review as product grows.
Requirements
- 5+ years product management or research operations, with ML/research team experience at tech/AI company.
- Technical depth to engage with safety researchers.
- Interest in AI safety and LLM deployment implications.
- Comfort in ambiguity with strong judgment.
- Ability to align researchers, engineers, product teams.
- Strong written communication for non-technical audiences.
Nice-to-Haves
- Experience with LLM evaluation, red-teaming, safety benchmarking.
- Familiarity with AI threats: prompt injection, jailbreaks, RAG poisoning, agentic misuse.
- Background in trust/safety, content policy, research operations.
- Building zero-to-one processes in research/safety.
- Exposure to agentic AI safety challenges.
Skills
LLM EvaluationRed-teamingSafety BenchmarkingPrompt InjectionJailbreaksRAGAgentic AIModel EvaluationsGuardrailsAI Safety Research