Staff Reinforcement Learning Research Engineer
Boston Dynamics3 months ago
Waltham, MA, USAStaff+
Base Salary
$155k - $200k/yr
Responsibilities
- Implement on-policy and off-policy reinforcement learning algorithms.
- Scale GPU-accelerated simulation to generate millions of samples per second.
- Develop sim-to-real transfer pipelines for policies deployed on physical robots.
- Integrate reinforcement learning with VLAs to fine-tune and distill large multimodal policies.
- Make robot policy deployment fast, easy, reliable, and reproducible.
- Build visualization tools that enable data-driven research.
- Own and develop company-wide RL tools supporting humanoid and quadruped robots.
Requirements
- An MS with 3+ years of experience, or a PhD, in machine learning, robotics, or a related field.
- Experience deploying policies on physical robots with attention to latency, robustness, and safety.
- Expertise with RSL-RL, CleanRL, RLlib, or Stable Baselines.
- Expertise with Isaac Lab, MuJoCo, MjWarp, or MjLab.
- Proficiency in PyTorch and/or JAX, plus inference runtimes such as ONNX, Triton, or TensorRT.
- Solid software fundamentals including Bazel, monorepos, Docker, and CI/CD.
- Experience building production-grade RL training pipelines.
- Deep knowledge of GPU-accelerated physics simulation.
- Experience applying RL to humanoid locomotion, whole-body control, or dexterous manipulation.
- Experience with sim-to-real transfer, domain randomization, or system identification.
- Experience with heterogeneous compute clusters and Kubernetes.
Benefits
- Medical, dental, and vision benefits.
- 401(k).
- Paid time off.
- Annual bonus structure.
- Ownership of the company-wide RL tools powering the company’s robots.
- Direct access to compute infrastructure for large-scale experiments.
Tech Stack
Categories
ML EngineeringRobotics
About Boston Dynamics
Boston Dynamics builds and sells mobile robots for industrial inspection, logistics, and research, including the Spot quadruped and the Stretch case-handling system. It generates revenue through hardware sales, software and service contracts, and fleet management tools for enterprise deployments. Founded in 1992 and headquartered in Waltham, Massachusetts, the company is majority-owned by Hyundai Motor Group and is developing humanoid robots for commercial tasks.