Research Engineer in Reinforcement Learning

Job Summary

Focusing on the development of algorithms and systems for training open models with reinforcement learning, the full-time Research Engineer in Reinforcement Learning will engage in algorithm research, scaling science, and enhancing RL frameworks, with the flexibility to work remotely or in-person in San Francisco.

Key responsibilities

• Develop and test reinforcement learning methods for various tasks and investigate stability issues in asynchronous RL
• Design experiments to analyze the impact of model size and training compute on RL performance and learning dynamics
• Extend and improve the performance of RL frameworks while building diagnostics for unstable runs and performance regressions

Required qualifications

• Demonstrated depth in RL algorithms, empirical research on model training, or distributed ML systems through research or open-source contributions
• Strong programming skills in Python and experience with PyTorch or a comparable ML framework
• Ability to design controlled experiments and interpret results effectively
• Experience in implementing and evaluating research ideas or systems changes with documented outcomes