Research Scientist, VLA Models - Atlas
Core
Architect, train, and deploy large-scale Vision-Language-Action (VLA) and behavior models for dexterous manipulation on the Atlas humanoid robot.
Role type
Senior IC research scientist (robotics/ML)
Builds
End-to-end VLA models, large-scale imitation learning pipelines, and hierarchical skill systems for mobile manipulation.
Domain
Robotics, Machine Learning, Humanoid Automation
Deliverable
production ML models
Required skills
Behavior cloning, imitation learning, diffusion policies, sequence modeling, multimodal learning, PyTorch, JAX, C++, Python
Preferred skills
RL integration, large demonstration dataset curation, top-tier conference publications, real-robot deployment experience
Technologies
PyTorch, JAX, C++, Python
Responsibilities
Architect and train end-to-end VLA models for mobile manipulation; build large-scale imitation learning pipelines from human/teleop/sim data; develop policies with few-shot generalization; create hierarchical behavior systems; integrate models into Atlas autonomy stack; deploy and debug models on physical hardware; write maintainable production code.
Seniority
Senior, hands-on IC
