Research Engineer / Scientist, Reinforcement Learning
Core
Adapt and improve leading open-weight models for legal work, specifically owning the post-training recipe across SFT and RL for long-horizon legal reasoning and agentic workflows.
Role type
Senior IC research engineer/scientist (reinforcement learning)
Builds
Post-trained LLMs for legal tasks, including reward functions, verifiers, and training runs
Domain
Legal technology + Large Language Models
Deliverable
production ML models
Required skills
RL for language models, SFT and preference optimization, distributed training, Python engineering, reward modeling, rigorous evaluation
Preferred skills
Publications in RL or reward modeling
Technologies
Python, SFT, RL, open-weight models
Responsibilities
Design and run post-training experiments across SFT and RL; Develop reward functions and verifiers for semi-verifiable tasks; Run and debug training at scale addressing divergence and instability; Build and use rigorous evals to track improvements; Design, implement, test, and debug code across the research stack
Seniority
Senior, hands-on IC
