What are the responsibilities and job description for the Research Scientist, Reinforcement Learning - Atlas position at Boston Dynamics?
At Boston Dynamics, we are pushing the boundaries of what advanced humanoid robots can do in the real world. The Atlas team is building next-generation whole-body mobile manipulation capabilities, and we are seeking a curious, driven Research Scientist to develop cutting-edge reinforcement learning (RL) solutions that run directly on our humanoid platforms.
In this role, you will design, train, and deploy RL policies that combine whole-body movement and dexterous manipulation to solve complex tasks in unstructured environments. You’ll work with a world-class team of roboticists and have rare, direct access to our physical Atlas robots and large-scale simulation infrastructure.
What You’ll Do
In this role, you will design, train, and deploy RL policies that combine whole-body movement and dexterous manipulation to solve complex tasks in unstructured environments. You’ll work with a world-class team of roboticists and have rare, direct access to our physical Atlas robots and large-scale simulation infrastructure.
What You’ll Do
- Design, implement, and train reinforcement learning algorithms for challenging whole-body mobile manipulation and bimanual manipulation tasks.
- Develop high-quality Python and C code that is tested, documented, and production-ready.
- Build and leverage high-fidelity simulation environments (e.g., Isaac Sim, MuJoCo) to validate RL policies before deploying on hardware.
- Integrate learned policies with Atlas’s control and software stack through close collaboration with controls and platform teams.
- Deploy, debug, and iterate policies directly on real Atlas hardware through hands-on experimentation.
- Participate in design reviews, experimental planning, and team-wide research direction.
- MS or PhD in Computer Science, Machine Learning, Robotics, or a related field.
- Strong experience training and deploying RL policies for complex behaviors in robots or simulated agents.
- Proficiency with modern ML frameworks (e.g., PyTorch, TensorFlow, RLlib).
- Strong foundations in algorithms, debugging, performance optimization, and robotics fundamentals (kinematics, dynamics).
- Excellent Python and C programming skills and experience contributing to production-scale software.
- PhD or equivalent research experience in reinforcement learning or robotic manipulation.
- Experience deploying RL policies on physical robots.
- Experience developing locomotion, bimanual manipulation, or whole-body control behaviors.
- Contributions to large software projects or open-source ML/robotics frameworks.
- Publications in top-tier robotics or ML conferences (e.g., CoRL, RSS, ICRA, NeurIPS).
- Direct access to cutting-edge humanoid robots and the infrastructure to run large-scale RL experiments.
- A highly collaborative, mission-driven team where your work has immediate impact.
- The opportunity to define state-of-the-art humanoid capabilities and shape the future of real-world robotics.