Humanoid Engineer, Manipulation

Job description

Job responsibilities

• Design, implement, and evaluate humanoid manipulation and loco-manipulation behaviors on real hardware
• Integrate perception, planning, control, grasping, whole-body coordination, and task execution into deployable robot workflows
• Run hardware experiments, analyze failures, and improve manipulation reliability across diverse objects, environments, and tasks
• Partner with system integration, hardware, field application, and testing teams to move capabilities from prototype to deployment
• Support teleoperation, data collection, and human-in-the-loop workflows for improving manipulation performance
• Build tools, metrics, and evaluation protocols for manipulation success, repeatability, failure recovery, and operator usability
• Debug cross-domain issues spanning software, sensors, actuators, end-effectors, calibration, timing, and field conditions

Job requirements

• MS or PhD in Robotics, Mechanical Engineering, Computer Science, or a related field preferred; BS considered with a demonstrated track record of hands-on robotics work across multiple physical systems — research projects, competition robotics, or internships with daily hardware exposure
• Hands-on experience with robotic manipulation, humanoids, mobile manipulation, dexterous hands, or contact-rich robotics — must include physical hardware; simulation-only backgrounds will not be considered
• Strong foundation in kinematics, dynamics, motion planning, control, and real robot experimentation
• Experience with C++, Python, ROS/ROS2, and Linux in a real robotics codebase
• Demonstrated ability to iterate quickly from experiment to working behavior on physical hardware; comfortable running daily hardware experiments, analyzing failures, and adapting approach in real-time
• Background appropriate for a junior-to-mid engineer; fresh MS and PhD graduates welcome
• Experience with humanoid platforms or contact-rich, dexterous manipulation systems — you’ve worked with robots that have hands, not just grippers
• Background in robot learning applied to physical hardware: imitation learning, reinforcement learning, or task and motion planning that you’ve validated on a real robot, not just in simulation
• You’ve taken a manipulation capability from prototype to reliable, repeatable field behavior — you know what it takes to close that gap and you’ve done it
• Track record of building evaluation frameworks for manipulation: test suites, metrics for success and failure, and the discipline to document and learn from what breaks