Software Engineer - Manipulation
Core
Develop and deploy learning-based manipulation policies for a humanoid robot to interact reliably with the physical world.
Role type
Senior IC robotics software engineer (learning-based manipulation)
Builds
Robotic manipulation policies for grasping, in-hand, and whole-arm control deployed on physical hardware
Domain
Robotics + Applied AI
Deliverable
production ML models
Required skills
robotic manipulation, RL fine-tuning, policy distillation, imitation learning, Python, C++, PyTorch/JAX, data pipeline engineering
Preferred skills
physics engines (IsaacSim, MuJoCo, Drake), generalist policies, teleoperation, tactile sensing, computer vision
Technologies
Python, C++, PyTorch, JAX, IsaacSim, MuJoCo, Drake
Responsibilities
Implement learning-based manipulation policies; adapt policies using post-training methods; build data pipelines for teleoperation and dataset curation; drive simulation-to-hardware transfer; define evaluation metrics and regression tests; collaborate with simulation, data, and hardware teams
Seniority
Senior, hands-on IC
