Pantelis R. Vlachas

Hi!

I’m Pantelis R. Vlachas, a Principal Machine Learning Scientist at Liom Health, where I work on machine learning for non-invasive health monitoring.

I am affiliated with ETH Zurich, where I work on generative modeling for spatial transcriptomics with Kalin Nonchev at Gunnar Rätsch’s Biomedical Informatics group.

At the same time, I am affiliated with the group of Prof. Eleni Chatzi, where I am exploring hypernetworks, inverse problems, parametric modeling and adaptive surrogate models of dynamical systems with online and real-time learning capabilities.

More broadly, my research lies at the intersection of machine learning, dynamical systems, and (lately) biology. It explores how ML methods enable the prediction of complex dynamics, inference of hidden properties, and discovery of underlying structure in systems ranging from turbulent flows to biological processes. Current focus areas include deep learning, probabilistic modeling, and scientific machine learning, with an emphasis on models that generalize to unseen regimes and adapt online.

Before that, I was leading the ML Research team at AI2C Technologies for four years, a spin-off from ETH working on novel machine learning solutions for fintech.

I completed my Ph.D. under the supervision of Prof. Petros Koumoutsakos in the CSE Lab in Zurich. There I worked on novel efficient methods for learning effective dynamics of complex systems. We demonstrated the effectiveness of the developed methods in a wide range of applications, from systems that exhibit spatiotemporal chaotic behavior and turbulent flows to small molecules. During my Ph.D. I also visited the John A. Paulson School Of Engineering And Applied Sciences in Harvard as an Associate Researcher.

I am always open to chat about cool projects, ideas, and collaborations.