The RoleAs a
Member of Clinical Staff, you'll help drive our research agenda towards advancing AI model capabilities on real-world healthcare tasks. You will build reinforcement learning environments, novel datasets and benchmarks, publish academic papers, and post-train models to push the SOTA while staying grounded in the demonstrated needs of our healthcare partners.
The clinical judgment you encode into our data decides whether a model gets a real clinical workflow right, or gets it confidently wrong.
This is a full-time role, required to be in-person in SF.What You'll Do- Design and build RL environments that capture real clinical work
- Find where frontier models fall short on real clinical workflows, and turn those gaps into environments, evals, and benchmarks
- Stress-test your own environments, hunting for shortcuts and reward hacks before a model finds them
- Write the rubrics and guidelines that capture expert clinical judgment, especially where the right answer is subjective or high-stakes
- Build gold-standard datasets, pilot new tasks yourself, and set the quality bar our clinician network works to
- Lead calibration with clinician raters and adjudicate the hardest cases
- Work with our ML team to turn training and evaluation needs into high-quality clinical data
- Write and publish papers in academic conferences
Who You Are- Must have one of the following medical degrees: MD, DO, MBBS, NP, or PA
- 2+ years of experience working in a real clinical environment
- You've done serious work in the human data space, at a frontier AI lab or a human data company (e.g., Scale AI, Surge AI, Mercor, Handshake AI, Turing). Not just completing tasks: you've designed evals, written the rubrics others worked from, led teams of clinical experts, or owned data that a frontier lab trained or evaluated on
- You've built RL environments, evals, or graders yourself, not just read about them
- You understand how data quality shapes model training, including preference data, rubric design, and rater agreement
- You understand healthcare data standards (e.g., EHR data structures, interoperability, HIPAA considerations)
- You're comfortable working in Python: you can read and modify environment and grader code, write scripts to build and analyze datasets, and dig into model outputs yourself. AI coding tools are welcome; you understand the code they produce well enough to debug and defend it
- You work in Git alongside engineers (branches, PRs, code review)
NICE TO HAVE- JavaScript/TypeScript, CI/CD, or AWS experience
- Experience shipping production software or digital health products
- Experience with clinical workflow analysis or quality improvement initiatives
Why UsIf you join our team, you will be joining a team with...
- Founding-level impact and ownership
- An opportunity to advance AI research in one of its most meaningful application areas
- Competitive compensation and meaningful equity
- Unlimited PTO
- Comprehensive health, dental, and vision coverage
- 401(k)
- Free lunch + dinner
