Geometric deep learning
Graph representations, expressive neural networks, and equivariant methods for structured data and biomolecular systems.
AI for science
George Dasoulas
Machine Learning Research Scientist
Isomorphic Labs / Switzerland
Developing machine learning methods
for scientific discovery.
01
About
I am a Machine Learning Research Scientist at Isomorphic Labs in Switzerland. My research spans geometric deep learning, multimodal learning, and foundation models for biology and drug discovery.
Previously, I was a Senior Machine Learning Research Scientist at Merck Research Laboratories, working on single-cell foundation models for target discovery, gene regulatory network inference, and multimodal learning for spatial transcriptomics and H&E imaging.
Before Merck, I was a Harvard Data Science Initiative Postdoctoral Research Fellow at Harvard University, appointed at Harvard Medical School and hosted by Marinka Zitnik’s Lab and Francesca Dominici’s Lab. During my Ph.D., I was a Doctoral Researcher in Machine Learning at the Mathematical and Algorithmic Sciences Lab, Huawei Paris Research Centre. I received my Ph.D. in Computer Science from École Polytechnique, working with Michalis Vazirgiannis, Aladin Virmaux, and Kevin Scaman.
02
Research
Learning from the structure and complexity of biological systems.
Foundation models for biology
Representation learning for single cells and proteins, connecting biological data with therapeutic discovery.
Multimodal learning
Combining complementary signals across molecular, spatial, and imaging modalities to understand biological function.
Recent work across biological foundation models and geometric learning.
Better Models, Faster Training: Sigmoid Attention for single-cell Foundation Models
STRAND: Sequence-Conditioned Transport for Single-Cell Perturbations
E(n) Equivariant Topological Neural Networks
ProCyon: A multimodal foundation model for protein phenotypes
Multimodal learning with graphs
Current
Isomorphic Labs
Machine Learning Research Scientist · Switzerland
Previously
Merck Research Laboratories
Senior Machine Learning Research Scientist · Cambridge, MA
2022–2025
Harvard University
HDSI Postdoctoral Research Fellow · Harvard Medical School
2018–2022
Huawei Paris Research Centre
Doctoral Researcher in Machine Learning · Mathematical and Algorithmic Sciences Lab
2019–2022
École Polytechnique
Ph.D. in Computer Science · DaSciM, LIX
2012–2018
National Technical University of Athens
B.Sc. & M.Sc. in Electrical and Computer Engineering
05
News & milestones
STRAND: Sequence-Conditioned Transport for Single-Cell Perturbations accepted to NeurIPS 2026.
Better Models, Faster Training: Sigmoid Attention for single-cell Foundation Models accepted to Transactions on Machine Learning Research; camera-ready version verified.
Joined Isomorphic Labs as a Machine Learning Research Scientist in Switzerland. Announcement ↗
TEDDY: A Family Of Foundation Models For Understanding Single Cell Biology — ICML GenBio Workshop.
E(n) Equivariant Topological Neural Networks — ICLR.
Wojcicki Troper postdoctoral fellowship, Harvard Data Science Initiative.