Researcher, Robot Learning
Core
Building generalist learning policies and self-improvement loops for Asimov, an open-source humanoid robot platform, enabling it to discover new skills and improve from real-world experience.
Role type
Senior Researcher, Robot Learning (Generalist Policies & Self-Improvement)
Builds
Generalist policies, imitation learning pipelines, and autonomous data collection loops for Asimov.
Domain
Robotics, Machine Learning, Humanoid Automation
Deliverable
production ML models
Required skills
Robot learning (imitation, reinforcement), Python, PyTorch, multi-GPU training, dataset engineering, sim-to-real transfer, evaluation design, open-source collaboration
Preferred skills
VLA models, diffusion policies, autonomous data collection, humanoid platforms, large-scale simulation (Isaac Lab, MuJoCo), top-tier conference publications
Technologies
Python, PyTorch, Isaac Lab, MuJoCo, mjlab, OpenVLA, Octo, LeRobot
Responsibilities
Own end-to-end generalist policy research from hypothesis to physical robot deployment; Train vision-language-action models and generalist policies; Build imitation learning pipelines on teleoperation data; Design self-improvement loops for autonomous data collection and learning from deployment; Combine RL, learning from demonstrations, and simulation to close the sim-to-real gap; Define evaluations for real hardware performance; Partner with Locomotion, Manipulation, and Platform teams; Publish and open-source research.
Seniority
Senior, hands-on IC
