Robotic Algorithm Engineer

We are building a next-generation humanoid robot platform with high-bandwidth torque-controlled joints and full-body actuation. Our short- to mid-term goal is to achieve robust and reliable locomotion in indoor service and industrial environments. As a Robotics Algorithm Engineer focused on Locomotion, you will work across simulation, learning-based control, state estimation, and real-robot deployment. This is a highly hands-on role requiring both strong implementation skills and the ability to debug complex real-world robotic behaviors. We are looking for engineers who not only implement algorithms, but also develop their own technical insights and adapt quickly in a rapidly evolving robotics landscape. Responsibilities Locomotion & Learning-Based Control Develop and deploy RL-based locomotion policies for humanoid robots Design training pipelines including domain randomization and sim-to-real transfer Improve policy robustness for indoor service and industrial use cases Analyze and debug failure modes from both simulation and real-world testing Full-Body Control & Modeling Work on whole-body control frameworks integrating learned policies Understand and leverage robot dynamics models for stability and contact reasoning Contribute to state estimation using IMU, joint encoders, and contact sensing Simulation & Tooling Build and maintain locomotion simulation environments (Mujoco / Isaac) Design training environments and reward shaping strategies Analyze the simulation-real gap and iterate on mitigation strategies Real Robot Deployment Deploy policies to hardware with torque-controlled, high-bandwidth actuators Perform real-robot tuning, debugging, and performance optimization Work closely with firmware, motor control, and hardware teams Qualifications Must Have 3+ years of experience in robotics, control, or locomotion-related roles Strong C++ and Python programming skills Experience applying reinforcement learning to robotics control problems Experience deploying algorithms on real robots (not simulation-only) Solid understanding of rigid body dynamics and feedback control Familiarity with state estimation for legged robots Experience working in Linux environments Nice to Have Experience with humanoid or legged robots Whole-Body Control or MPC exposure Mujoco / Isaac Gym / Isaac Sim experience Experience addressing sim-to-real transfer challenges Familiarity with Pinocchio, CasADi, or similar tools CUDA or large-scale RL training experience ROS2 experience Work Mode On-site required Up to 10% of travel