Researcher, Robot Learning

Menlo • Palo Alto, CA

2d

About The Position

Menlo Research is an applied R&D lab building Asimov, an open-source humanoid robot platform and its full software stack. The Researcher, Robot Learning will join the Brain team and own methods for turning data into general-purpose behavior on Asimov, including generalist policies, training data, and systems that enable the robot to improve from its own experience. The work spans research through deployment: train on real-robot data, run policies on physical hardware, and open-source work where possible. The role collaborates with teams across World Models, Locomotion, Manipulation, and Platform to integrate research into the shipping stack.

Requirements

• Deep hands-on experience training learning-based robot policies through imitation learning, reinforcement learning, or both.
• Experience taking a learned policy from research code to a physical robot and addressing deployment challenges.
• Fluency in Python and PyTorch.
• Comfort training at scale on multi-GPU clusters.
• Understanding of how dataset composition, quality, and diversity affect policy capabilities.
• A PhD or equivalent research experience in robot learning, machine learning, or a related field, or a portfolio that demonstrates relevant ability.
• Clear communication through papers, blogs, talks, or well-documented code.
• Use of AI as a collaborator in research.

Nice To Haves

• Experience with VLA models or robot foundation models such as OpenVLA, Octo, or π0.
• Experience with diffusion policies, flow matching, or action chunking.
• Experience with autonomous data collection, reward learning, or continual learning on real robots.
• Experience with humanoids or other high-DoF, whole-body platforms.
• Familiarity with large-scale simulation in Isaac Lab, MuJoCo, or mjlab.
• Published work at CoRL, RSS, ICRA, NeurIPS, or similar venues.
• Contributions to open-source robot learning libraries such as LeRobot.

Responsibilities

• Own Asimov’s generalist policy research end to end, from hypotheses and datasets to trained policies running on the physical robot.
• Train and post-train vision-language-action models and other generalist policies that follow instructions and transfer across tasks and environments.
• Build imitation-learning pipelines using teleoperation and demonstration data, and assess which data improves performance.
• Design a self-improvement loop covering autonomous data collection, success detection, and learning from deployment.
• Combine learning from demonstrations, reinforcement learning, and simulation to address the sim-to-real gap at scale.
• Define evaluations that determine whether policies improve on real hardware, not only on benchmarks.
• Partner with World Models, Locomotion and Manipulation, and Platform so research is incorporated into the shipping stack and supports agents using Asimov through its SDK.
• Publish, open-source, and share research with the community building on Asimov.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: Ph.D. or professional degree