- RL environments, simulated tool use, or long-horizon agent benchmarks.
- Synthetic data generation, human review workflows, or dataset quality systems.
- Reward design, model-based graders, and validation against expert judgments.
- Building evaluation or training data products for AI labs.
- Healthcare data, clinical reasoning evaluations, or longitudinal patient workflows.
Healthcare experience is welcome, but it isn’t required. Curiosity about the domain and a willingness to work closely with clinicians matter.
You’ll help define how we measure progress toward AI that can support care over time. Your work will make it possible to investigate failures, compare approaches, and build training signals grounded in realistic tasks.
You’ll own both the experimental design and the engineering needed to make it useful, with room to shape a new area of work at Nolla.
We’d ideally like you to be based in New York City.