Who we are, and how to join.

Edward Johns. Co-Founder and CEO.
Professor of Robot Learning at Imperial College London. In 2020, started building a research group dedicated to one-shot robot learning, and developed many of the early methods in this space. Brought that same team and six years of research and technology into Machenta.

Vitalis Vosylius. Co-Founder and CTO.
PhD from Imperial College London. Built the first generalist in-context learning models for robotics trained entirely in simulation, Implicit Graph Alignment (CoRL 2023) and Instant Policy (ICLR 2025). Awarded Imperial’s PhD Prize for Outstanding Achievement in Robotics.

We are a team of researchers and engineers, with origins in the Robot Learning Lab at Imperial College London and beyond. In 2020, we established our vision that one day, robots would be able to learn like humans — one demonstration, instant learning — and we focussed on this one problem for six years. Our research took us from our 2021 paper on self-supervised learning that started it all, through to the first work on generalist in-context learning in robotics in 2023, and our work on trajectory alignment for one-shot learning also in 2023. In 2025, we then had two major breakthroughs which triggered the founding of Machenta. Firstly, Instant Policy, which showed in-context learning was possible by pre-training only in simulation, winning awards from Microsoft and Amazon. Secondly, Learning a Thousand Tasks in a Day, which showed how one-shot learning at scale leads to unprecedented task learning efficiency and generalisation, making it to the front cover of Science Robotics. Every author across these five papers is now building Machenta, alongside researchers from DeepMind, Dyson, Huawei, and Stability AI.