Safety Jevroni
TypeSafe AI is an AI lab building machine-native intelligence infrastructure for software automation. Its flagship System One model, Jev, provides typed decisions, probabilities, and confidence scores through an API and SDKs for developers building automated workflows.
Maintainer signals as of 9/25/2026
Projects
About TypeSafe AI
TypeSafe AI develops System One models for fast, structured decisions that software can consume directly rather than generated text. Its flagship model, Jev, evaluates supplied state against typed Choice, Score, and Noul questions, returning constrained results, probabilities, and confidence values. The company provides an API, Python and JavaScript SDKs, developer documentation, a console playground, and tools intended for developers building reliable automated software workflows.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will build evaluations and investigate model behavior that concerns you. You will raise questions in meetings, document your observations, and prepare a departure post. During the six-week contract, you will get set up, request access, write an internal memo, and decide whether to return.
Requirements
- Strong research background in AI safety, alignment, interpretability, or model evaluation
- Experience raising concerns calmly and in writing
- Ability to avoid discussing specifics
- Sincere belief that you can change things from the inside
Responsibilities
- Build evaluations and investigate concerning model behavior
- Raise questions about whether proposed actions should be taken
- Document initial observations and concerns
- Write a departure post
- Get set up and meet the team
- Request additional access
- Write a long internal memo
Hiring Process
Submit a draft of your eventual resignation post; specifics may be omitted.