The team

The people behind Desupervised

A research-led company with a small, focused team. We publish our work, open-source our engine, and ship production models under our own names.

The Desupervised team

Founders

Dr. Michael Green

Dr. Michael Green

CEO, Founder

PhD, Theoretical Physics

Michael holds a doctorate in theoretical physics and has spent 25 years building neural networks. He founded Desupervised in 2018 and as CEO sets the company's direction: co-developing uncertainty-aware AI solutions with companies that can't afford overconfident predictions. He is a vocal advocate for uncertainty quantification: a model that cannot say what it does not know should not be trusted to make decisions. Before tech, he led a team in Kustjägarna, the Swedish Navy's coastal ranger commando unit.

Bayesian inferenceProbabilistic programmingUncertainty-aware deep learningApplied ML
Johan Gudmundsson

Johan Gudmundsson

CTO, Co-Founder

MSc, Applied Mathematics

Johan co-founded Desupervised in 2018 with Michael Green. As CTO he leads engineering and is the lead author of Borch, the open-source probabilistic programming language underneath everything Desupervised builds. His focus is shipping co-developed, uncertainty-aware AI into production with clients: from one-line conversion of PyTorch models to Bayesian equivalents, through to equivariant graph neural networks for drug discovery.

Probabilistic programmingDeep learning infrastructureBorchEquivariant graph neural networks
Jens Skoustrup-Jacobsen

Jens Skoustrup-Jacobsen

CFO, Co-Founder

MBA · Accountant

Jens co-founded Desupervised in 2018 and runs the financial and operational side of the company. As CFO he handles funding, contracts, and the day-to-day machinery that lets the research and engineering teams ship.

FinanceOperationsStrategy

Advisory board

Dr. Mattias Ohlsson

Dr. Mattias Ohlsson

Scientific Advisor

PhD, Theoretical Physics

Mattias serves as scientific advisor to Desupervised, bringing decades of academic and applied research in machine learning and Bayesian methods to the company's research direction.

Machine learningComputational physicsBayesian methods
Dr. Andreas L Norlin

Dr. Andreas L Norlin

Scientific Advisor

PhD

Andreas serves as scientific advisor to Desupervised, contributing strategic guidance on the company's research direction and applied programmes.

StrategyGovernanceApplied research

Research-led, production-hardened

We publish our work openly. See publications, explore Borch, or read the blog.

Ready for the next step?

Tell us the decision you need to make with confidence. We'll tell you honestly whether a model can help.