Emmanuel
de Bézenac

Permanent researcher at INRIA Paris, working on deep learning for weather forecasting and generative models.

ARCHES · INRIA Paris · Scholar · LinkedIn
About

I am a permanent researcher (ISFP) at INRIA Paris in the ARCHES team, led by Claire Monteleoni.

My current work focuses on building AI weather forecasting models. I am also broadly interested in generative models — including diffusion models, flow matching, and consistency models — and in developing deep learning methods that leverage the structure of physical systems.

Previously, I was a postdoctoral researcher at ETH Zürich with Siddhartha Mishra, where I worked on neural PDE solvers, scaling them to large-scale problems, and investigating the theoretical foundations of these approaches — including variable-resolution operators and the fundamental difficulties of physics-informed deep learning.

I obtained my PhD from Sorbonne Université, supervised by Patrick Gallinari, where I have been working since 2017 on modeling physical processes with deep learning, developing models that embed prior scientific knowledge and PDEs.

Weather Forecasting Generative Models Neural PDE Solvers Physics-Informed ML
Selected Publications

All publications →

Selected Talks
Open Positions

Our team may have openings for postdocs, PhD students, and M2 research interns. If the topics on this page resonate with you and you're genuinely excited about this line of research, feel free to reach out.

Get in touch
Blog

Coming soon — in English and French.

Contact