Julie Keisler

I am currently a researcher in the ARCHES team from INRIA Paris. My research focuses on AI-driven methods for statistical downscaling. I am also interested in automated machine learning, particularly neural architecture search, and the use of AI methods for the energy transition.

I completed my PhD in January 2025 under the supervision of Claire Monteleoni (INRIA Paris and the University of Colorado Boulder), El-Ghazali Talbi (INRIA Lille and Université de Lille), Margaux Brégère (EDF R&D and Sorbonne Université), Gilles Cabriel and Sandra Claudel. The thesis is entitled Automated Deep Learning: algorithms and software for energy sustainability. The manuscript is available here, the defense can be watched here and the slides found here.

I developed a Python package called DRAGON to optimise deep neural networks and, more recently, generalised additive models. The code is available on my GitHub. If you would like to test it, please get in touch!