Founding Research Scientist

Founding Research Scientist 

What would you work on if you could help define the research agenda from day one?

We’re hiring a Founding Research Scientist to join an ambitious UK Physical AI startup building the models that will enable robots to understand, learn and act in the real world.

This is an opportunity to pursue difficult, open-ended research problems without being confined to an established roadmap. You’ll develop new ideas, test them on real robotic systems, and turn promising research into technology that works outside the lab.

What’s in it for you?

• Founding role with significant equity
• Help define the research agenda from the ground up
• Work at the frontier of robot learning, vision-language-action models and world models
• Design and run experiments on real robotic hardware
• Publish meaningful research while building technology with real-world impact
• Work closely with a small, highly technical founding team
• Have genuine ownership over which problems are worth solving and how to approach them

What you’ll work on

• Lead research across robot learning, world models and VLA systems
• Develop new approaches to model architectures, training and data collection
• Design rigorous evaluation methods for Physical AI systems
• Take ideas from first principles and papers through to prototypes and real-world robotic experiments
• Investigate what enables models to generalise across tasks, environments and embodiments
• Publish research and help establish the company’s long-term technical direction

What you’ll bring

• Strong research experience across machine learning, robotics or computer vision
• Deep understanding of modern deep learning methods
• Excellent mathematical, experimental and problem-solving instincts
• Hands-on experience training and evaluating ML models
• The ability to form hypotheses, design experiments and learn quickly from results
• Curiosity for problems where the answers — and sometimes even the right questions — aren’t obvious

Publications, benchmarks and open-source contributions are valuable signals, but they aren’t the deciding factor.

What matters most is the quality of your research thinking, the problems you choose to pursue, and your ability to turn ideas into experiments that teach us something new.