At UMA, we’re pushing the frontier of physical AI by teaching robots to understand, plan, and reason in the physical world. A critical part of this effort is building world models that can predict the outcomes of actions before they are executed on a robot.
As a Research Scientist on the World Models team, you’ll own a piece of that frontier and drive it end to end: from ideas to real-world capabilities for humanoid robots deployed in the industry. You’ll have the autonomy of a research lab and the urgency of a startup shipping real products, working alongside people who built robots at Tesla Optimus, DeepMind, Google Brain and LeRobot at HuggingFace.
Key Responsibilities:
Design, run, and analyze original experiments on world models, planning, and reasoning, in simulation and real-world benchmarks.
Own the stack from data processing, model training to evaluation on real robots.
Push the frontier on self-supervised pretraining across video and action data for real-world manipulation and locomotion.
Take your research all the way onto hardware, integrating with the control stack, debugging closed-loop behaviour, and iterating against the messiness of the real world.
Document and present your work, and mentor the engineers who help bring it to life.
A PhD (or equivalent industry research track) in robotics, representation learning, world models, self-supervised learning, RL/planning, or video modeling with a track record of papers and/or shipped systems.
Real depth in the relevant stack: large-scale training, efficient inference, fast experimentation.
The rare combination of research originality and engineering pragmatism with end to end ownership.
Experience with latent world models like JEPA is highly valued; the ability to learn fast and bet well matters more than a specific paper on your CV.
Bonus: You’ve worked in early-stage startups or the core teams of big companies building humanoids or highly-dexterous robots.
UMA is an inclusive workplace that values exceptional builders over perfect pedigrees. Whatever your background, identity, or journey, if you don't meet every criterion but believe you can have an outsized impact here, we strongly encourage you to apply.
Skills Required
- PhD or equivalent industry research track in robotics, representation learning, world models, self-supervised learning, RL/planning, or video modeling
- Track record of papers and/or shipped systems
- Experience with large-scale model training, efficient inference, and fast experimentation
- Ability to take research end-to-end onto hardware and integrate with robot control stacks
- Research originality combined with engineering pragmatism and end-to-end ownership
- Experience with latent world models (eg. JEPA)
- Experience at early-stage startups or core teams building humanoid or highly-dexterous robots
What We Do
We build general-purpose mobile and humanoid robots capable of human-level dexterity and understanding of the physical world. It will enable people to focus on what truly matters in their lives. We are based in Paris, FR. Join us: https://app.dover.com/jobs/uma
