Maxime BOUTON
PhD in Signal and Image Processing
I am a R&D Engineer and Researcher specializing in High-Performance Computing (HPC) and Scientific Computing. I recently completed my PhD focused on scalable Bayesian inference and Plug-and-Play MCMC algorithms for high-dimensional inverse problems.
During my research, I bridged the gap between advanced mathematics and production-grade software by designing and developing CARDS, an open-source Python library for distributed sampling on multi-GPU clusters using PyTorch and MPI. Driven by code efficiency and structural depth, I enjoy optimizing memory footprints, accelerating computation times, and deploying cutting-edge AI models into complex, large-scale industrial architectures (such as numerical simulation or game engines).
I started my PhD in October 2022 under the supervision of Pierre-Antoine THOUVENIN (Maître de Conférences) and Pierre CHAINAIS (Professeur), and defended in December 2025.
PhD Title: "Distributed Plug-and-Play MCMC algorithms for large Bayesian inference"
Research Interests
Computational Imaging & Inverse Problems
Development of advanced methods for solving large-scale inverse problems in imaging, with focus on Bayesian approaches and uncertainty quantification.
Markov Chain Monte Carlo (MCMC)
Design and implementation of efficient sampling algorithms for high-dimensional Bayesian inference, including Plug-and-Play approaches with learned priors.
Machine Learning & Neural Networks
Integration of deep learning techniques, particularly denoisers and neural network priors, within Bayesian inference frameworks for imaging applications.
Distributed & High-Performance Computing
Scalable implementation of sampling algorithms on multi-GPU architectures to tackle large-dimensional problems in computational imaging.