About this role
As a Research Engineer/Scientist specializing in Reinforcement Learning, you will be instrumental in adapting and enhancing cutting-edge open-weight models for complex legal reasoning and agentic workflows. You will own the post-training process, including reward design and verification for challenging tasks, and drive research from experimentation to customer-facing models within a fast-paced, high-impact environment.
What we are looking for 6
- Own research agenda for long-horizon legal reasoning and agentic workflows
- Design and execute SFT and RL post-training experiments
- Develop reward functions and verifiers for semi-verifiable tasks
- Run and debug large-scale training, addressing instability and reward hacking
- Build and utilize rigorous evaluations to track model improvements
- Hands-on experience post-training LLMs with SFT and RL