Research / Software Engineer - Humanoid Whole Body Learning

Job description

Job responsibilities

• Develop and train whole-body loco-manipulation policies for humanoid robots.
• Deploy trained policies to physical humanoids and integrate them into our production software stack.
• Advance our motion retargeting pipeline, transforming human motion into physically plausible, robot-executable behaviors that interact with diverse environments and objects.
• Reduce the sim-to-real gap by improving simulation fidelity and closing the real-to-sim loop, using real-world robot data to calibrate and refine our simulators.
• Build automated evaluation and validation systems that make it faster and more reliable to move policies from simulation onto physical robots.
• Improve the performance and scalability of our robot-learning infrastructure, enabling faster experimentation and policy iteration.
• Work closely with researchers and engineers across perception, learning, simulation, and hardware to turn new ideas into deployed robot capabilities.

Job requirements

• BS, MS, or PhD in Robotics, Computer Science, Machine Learning, Engineering, or a related field, or equivalent experience.
• Experience with reinforcement learning, imitation learning, generative models, or other learning-based approaches for robotics.
• Strong understanding of robotics fundamentals such as kinematics, dynamics, control, and physical interaction.
• Experience developing and evaluating robotic systems in simulation and/or on physical hardware.
• Ability to move comfortably between research experimentation and production-quality engineering.