Thesis Defense Scheduled – September 5, 2025

I’m happy to announce that the date for my PhD defense has been decided for Friday the 5th of September 2025. My thesis titled

“Real-Time Analysis of Unstructured Data with Machine Learning on Heterogeneous Architectures”

describes my work, along my collaborators, in developing ETX4VELO, a Graph Neural Network-based pipeline for real-time track reconstruction at 40 MHz inside the LHCb first-level trigger. The pipeline was developed in PyTorch, implemented end to end on GPUs using the C++ CUDA framework by Nvidia, and partially implemented on FPGAs using the translation framework HLS4ML, which transforms PyTorch/Keras code to firmware for low-latency, high-throughput inference on FPGAs. For more information on the project you can also see my Projects page. The thesis was conducted under the co-supervision of Vava Gligorov (LPNHE/CERN) and Bertrand Granado (LIP6/Sorbonne Université).

More updates and a link to the manuscript will follow soon.