Safety Jevroni (Contract, 6 weeks)
TypeSafe AI · San Francisco Office
Frontier labonsite
At a glance
Highlights
- Build evals and investigate concerning behavior in frontier System One Models.
- Probe trustworthiness of Jev to make AI dependable enough for production.
- Rebuild the LLM stack from first principles for reliable intelligence.
Ideal candidate
A principled AI safety researcher who cares deeply about responsible deployment and is motivated to probe hard questions from inside a frontier lab.
Requirements
- Strong research background in AI safety, alignment, interpretability, or model evaluation
- Experience raising concerns calmly, then in writing
- Ability to say you can't discuss specifics
- Sincere belief you can change things from the inside
- hiring process
- To apply, submit a draft of your eventual resignation post
Lab profile
About TypeSafe AI
Building System One Models that combine LLM reasoning with code reliability for production AI.
What they build
- Jev
- System One Models
Known for
- Production-grade trustworthy AI
- LLM reasoning with code reliability
- First-principles LLM stack rebuild
Key facts
- Small, fast team
Posted · Last seen