Aymeric Delefosse

I’m a PhD candidate at Inria, working in the ARCHES team under the supervision of Anastase Charantonis and Dominique Béréziat. My research lies at the intersection of climate science and machine learning, where I focus on reducing uncertainties in learned circulation models.

More specifically, I’m currently exploring how generative AI can be used for downscaling, i.e. turning coarse climate simulations into finer details needed for local-scale analysis.

I completed my Master’s in Computer Science at Sorbonne Université, where I did my thesis in collaboration between LOCEAN and ISIR. There, I worked on data imputation methods for oceanographic variables, especially phytoplankton, which deepened my interest in applying machine learning to environmental questions.

I’ve always been drawn to the intersection of data, complex systems, and our changing environment. My background spans statistics, applied mathematics, and computer science, and I’m motivated by the potential of artificial intelligence not just to predict, but to understand the physical systems we depend on.

When I’m not debugging code or reading papers, you’ll likely find me hiking or exploring museums.