Robot Learning Researcher — Real-World Humanoid AI

Menlo; Palo Alto (CA); Presencial; USD 150 000 - 230 000; Tempo integral

Resumo da oferta

Menlo Research is hiring a Researcher in Robot Learning on the Brain team. You will own end-to-end methods turning data into general-purpose behavior for Asimov, training policies on real robot data and shipping to hardware.

You will develop and evaluate imitation and reinforcement learning pipelines, design autonomous data collection loops, and push toward sim-to-real performance improvements at scale.

Qualificações

• Deep hands-on experience training learning-based robot policies (imitation or RL)
• Experience taking a learned policy from research to a real robot and handling deployment issues

Responsabilidades

• Own generalist policy research end to end for Asimov, from hypothesis and data to trained policy on robot
• Train and post-train vision-language-action models and other generalist policies
• Build imitation learning pipelines on teleoperation and demonstration data and evaluate impact on performance
• Design self-improvement loop with autonomous data collection, success detection, and learning from deployment
• Combine learning from demonstrations, reinforcement learning, and simulation to close the sim-to-real gap at scale
• Define evaluations to measure policy performance on real hardware, not just benchmarks
• Collaborate with World Models, Locomotion/Manipulation, and Platform teams to ship integrated capabilities

Conhecimentos

Generalist policy research

Imitation learning

Reinforcement learning

Python

PyTorch

Experimentation at scale

Formação académica

PhD in robot learning or ML

Ferramentas

PyTorch

ROS

Descrição da oferta de emprego

Menlo Research is hiring a Researcher in Robot Learning on the Brain team. You will own end-to-end methods turning data into general-purpose behavior for Asimov, training policies on real robot data and shipping to hardware.

You will develop and evaluate imitation and reinforcement learning pipelines, design autonomous data collection loops, and push toward sim-to-real performance improvements at scale.