World Models Engineer for Humanoid Robotics

Menlo; San Francisco (CA); On-site; USD 150,000 - 210,000; Full time

Job summary

Menlo Research is hiring a Researcher to advance world models at the core of Asimov, our open-source humanoid robot platform. You will work at the intersection of self-supervised representation learning, predictive architectures, and embodied control, collaborating with platform, firmware, and hardware teams.

You will design, train, and evaluate models, own data pipelines end to end, and help bridge research to real hardware deployments, contributing to sim-to-real transfer and multi-modal

Qualifications

• Proven modeling track record with evaluated results.
• Understanding JEPA and world models in robotics.
• Breadth across vision, language, or sensor modalities.

Responsibilities

• Design, train, and evaluate world models for action prediction.
• Advance self-supervised learning for visual and sensor reps.
• Prototype predictive architectures and compare with JEPA objectives.
• Own data pipeline end-to-end: curation, tooling, scaling.
• Integrate research with platform and firmware teams for real hardware.
• Publish or open-source work to strengthen the field and team.

Skills

Modeling experience

JEPA fluency

Multimodal sensor fusion

PyTorch

JAX

Distributed training

Robotics ML

Education

PhD or equivalent research

Tools

PyTorch

JAX

DIffusion/Generative models

Job description

Menlo Research is hiring a Researcher to advance world models at the core of Asimov, our open-source humanoid robot platform. You will work at the intersection of self-supervised representation learning, predictive architectures, and embodied control, collaborating with platform, firmware, and hardware teams.

You will design, train, and evaluate models, own data pipelines end to end, and help bridge research to real hardware deployments, contributing to sim-to-real transfer and multi-modal