Manipulation Capabilities Engineer
Core
Develop and deploy manipulation policies for humanoid robots using deep learning and reinforcement learning to enable real-world interaction.
Role type
Senior IC manipulation capabilities engineer (robotics & deep learning)
Builds
HMND-01 Alpha humanoid robot manipulation capabilities
Domain
Robotics, Deep Learning, AI
Deliverable
production ML models
Required skills
robot hardware experience, low-level control stack, neural network post-training, data preprocessing, RL training, PyTorch or JAX, distributed training, software engineering practices
Preferred skills
VLA model training (autoregressive, diffusion, flow-matching), OpenVLA, Physical Intelligence (π) models, RL in robotics, top-tier conference publications
Technologies
PyTorch, JAX, OpenVLA, Physical Intelligence (π)
Responsibilities
Post-train manipulation policies via behavior cloning and RL; develop data preprocessing strategies; collaborate with simulation teams on digital twin RL training; drive data collection activities for specific capabilities; expand observation and action spaces; partner with teleoperations to improve motion smoothness; interface with hardware design team
Seniority
Senior, hands-on IC
