Denis Barthou

I am full professor in Computer Science, at Bordeaux INP/ENSEIRB-MATMECA. I graduated from ENS Lyon and earned a PhD from the University of Versailles–Saint-Quentin in 1998, where I worked on dependency analysis in the polyhedral model. I then became an assistant professor at the same university. In 2009, I joined Bordeaux INP as a full professor and became part of Inria’s Runtime team. In 2017, I created and led Inria’s Storm team while also overseeing the Computer Science program at the ENSEIRB-MATMECA engineering school of Bordeaux INP. From 2023 to mid-2025, I was on leave to direct Huawei Paris’s Distributed and Parallel Research Lab. After returning from this leave, I joined Inria’s Topal team.
Over the past two years, I’ve increasingly focused on how to scale AI systems, especially open-source large language models. I’ve explored parallelization techniques for both training and inference, and how these models behave when deployed on large cloud infrastructures built around modern AI accelerators. Along the way, I’ve gained substantial hands-on experience with a variety of heterogeneous hardware platforms—from the familiar NVIDIA GPU ecosystem to architectures like Huawei’s Ascend NPUs, as well as multicore ARM and Intel processors. More broadly, my research has always revolved around making complex applications run faster and more efficiently. I’m interested in the full stack: optimization, parallel and distributed methods, and the design of compilation and runtime systems for AI and high-performance computing. These themes have accompanied me throughout my career, even as the computing landscape continues to evolve.