Andrey Malinin
I work on artificial intelligence for science.
About
I’m a research scientist working on generative AI for science — to me, the most exciting field to be in right now. In just a few years, AI has gone from a convenience to something that can predict how proteins fold, prove hard theorems, design novel materials and propose new hypotheses. I believe AI could bring about a new era of scientific discovery and progress on the scale of the Industrial Revolution, and I am eager to help build that future.
Right now I’m a Member of Technical Staff at CuspAI, designing new materials. Prior to CuspAI I worked on AI-driven drug discovery at Isomorphic Labs, and before that at Yandex Research, where I founded the Shifts Project. I hold a PhD in machine learning from the University of Cambridge, supervised by Professor Mark Gales. A running theme throughout my research has been out-of-distribution generalisation — getting models to work reliably on novel inputs, which is central to applying AI in science.
In my spare time, I’m an analogue photography enthusiast and a lover of all things vintage.
Now
Now · Aug 2026At CuspAI, designing new materials with artificial intelligence. Reading about generative models, machine-learning force fields, and DFT. London.
Where I’ve worked
CuspAI · Member of Technical Staff
Generative models for materials discovery.
Isomorphic Labs · Staff Research Scientist
Biomolecular interaction prediction for AI-driven drug discovery.
Yandex Research · Principal Research Scientist
Founded the Shifts Project; robustness and uncertainty at industrial scale.
University of Cambridge · PhD, Information Engineering
Uncertainty estimation in deep learning, applied to spoken language assessment.
Selected publications
- 01
Shifts: A Dataset of Real Distributional Shift Across Multiple Large-Scale Tasks
NeurIPS 2021 · Datasets & Benchmarks · Repo
- 03
Ensemble Distribution Distillation
ICLR 2020
- 05
Predictive Uncertainty Estimation via Prior Networks
NeurIPS 2018
Full list → Google Scholar