Team · Edinburgh, Scotland
A team with roots in Scotland.
We are in stealth, which hides what we are building and not who is building it. Both papers the company rests on are public, and the people who wrote them are below.
Djordje Mihajlovic
Cofounder
PhD Mathematics, University of Edinburgh
TAP Lab (Biophysics) · Renaissance Philanthropy Fellow
His work here is on the internals of biological foundation models for proteins and RNA, read through geometry, topology and biophysics: describing what a model has learned using mathematics first, capturing the geometry of a protein backbone and the topology of an RNA fold.
He is a PhD student in mathematics at Edinburgh, jointly supervised by Davide Michieletto in the Topologically Active Polymers lab and by Tudor Dimofte, using machine learning to investigate knot theory across topology and geometry. Alongside it he worked with the Renaissance Philanthropy AI for Math Fund on agent evaluation over mathematical datasets — measuring what a model can do from the outside, and seeing where that measurement stops. Before both, a masters in theoretical physics.
Siddharth Setlur
Cofounder
PhD Mathematics, University of Edinburgh
His work here applies the framework of susceptibilities to structural inference of biological foundation models. That would be a significant step towards a mathematical auditing of model internals for harmful capabilities, beyond what naive evaluations reach.
He is a PhD student in mathematics at the University of Edinburgh, jointly supervised by Darrick Lee and Tudor Dimofte, and an Iliad fellow, where his research is on theoretical approaches to AI safety. Before Edinburgh he was at ETH Zürich, where he obtained an MSc and a BSc in Mathematics. He has worked with Inria, the French National Institute for Research in Digital Science and Technology.
Max Condouret
Cofounder
MSc Applied Mathematics, ETH Zürich
Visiting graduate researcher, MGH and Harvard Medical School
His focus here is to fundamentally understand medical-grade image foundation models: decoding their inner workings by applying advanced mathematics to parse exactly how these systems learn visual features.
He is reading for a master’s in Applied Mathematics at ETH Zürich, with a background in machine learning and statistics applied to biology. His research track record includes a master’s thesis on image-to-omics predictions under Prof. Jian Shu at Harvard Medical School, and applied statistics research for single-cell biology under Prof. Andreas Krause at the ETH AI Center.
Scientific advisors
Academic advisors across biology and machine learning interpretability.
Working with a protein model? See what we offer.