Sr. Research Scientist - Agent Flywheel

As a Research Scientist on the Agent Flywheel team at Databricks, you will build the context layer for recursive self-improvement and autofix loops for agents. Your primary focus will be on turning raw production session traces from Databricks' data agents into online signals agents can optimize against.

You will research how to convert messy, real customer interactions into deterministic metrics, calibrated LLM judges, and failure clusters, and close the loop so those signals drive training and agent improvement.

Your responsibilities will include:

• Turning raw production session traces from Genie Code and Genie One into online signals agents can optimize against.
• Training and calibrating LLM judges so model feedback becomes reliable, repeatable signal, including calibration against expert human ratings.
• Converting multi-turn agent trajectories into datasets and reward signals for post-training, closing the loop between production signals, evaluation, and agent improvement.
• Designing benchmarks and RL environments from real customer workloads that probe long-horizon, tool-using agent behavior.
• Advancing evaluation research: online evals on live traffic, automated raters, verifiable rewards, and measurement that keeps pace with agents that act.
• Keeping up with the latest developments in deep learning and creating new techniques that go beyond the state of the art.

We are looking for someone with expertise in debugging and diagnosing failures in complex, stochastic, multi-component systems, as well as a deep instinct for designing metrics and evals that track what actually matters.

The ideal candidate will have solid AI research and software engineering skills, strong communication skills, and a drive to get their work into the hands of real users.

The pay range for this role is $164,000-$235,000 USD.