Menlo Research
Researcher, World Models
Palo Alto +1 · Hybrid · Full-time · Posted Jul 10, 2026
- Location
- Palo Alto
Singapore - Education
- PhD or equivalent research experience in ML, robotics, or computer vision is a plus, not required with a strong portfolio.
About the job
Develop and integrate predictive world models for Asimov humanoid robots, enabling them to perceive, predict, and act across sensory modalities.
What you'll do
- Design, train, and evaluate world models and self-supervised representations for embodied prediction and planning.
- Own experiment data pipelines and benchmark predictive and generative architectures.
- Integrate research with platform, firmware, and software teams; contribute to sim-to-real transfer, inverse dynamics, and sensor fusion.
What they ask for
- Proven track record training models and evaluating them rigorously.
- Understanding of JEPA and ability to compare it with alternative approaches.
- Depth in at least one sensory domain, such as vision, audio, or natural language.
- Ability to implement, integrate, and ship models, with strong data skills.
Nice to have
- Publications at NeurIPS, ICML, ICLR, CoRL, or RSS, or comparable-citation arXiv work.
- Hands-on hardware or robotics interest or experience; PhD or equivalent research experience is also noted as a plus, but not required with a strong portfolio.
Tools
V-JEPA, I-JEPA, LeJEPA, MJEPA, diffusion, DiT, flow matching, PyTorch
AI focus
Research and build self-supervised and predictive world models for humanoid robotics, including embodied prediction, planning, and integration on real hardware.
Model-written summary · the original posting is authoritative