François-david Collin
Ingénieur De Recherche - Intelligence Artificielle & Calcul Scientifique (LIRMM CNRS)
Montpellier, Occitania, France
Summary
🤩
Rockstar🎓
Top SchoolFrançois-david Collin is a research engineer specializing in AI and scientific computing at LIRMM/CNRS with 11+ years delivering research-grade software, numerical models, and rapid prototyping for R&D and artistic embedded systems. He blends a solid theoretical computer science and renewed mathematics background with broad operational experience as developer, sysadmin, network architect and security analyst, enabling end-to-end production of reproducible scientific tools. Notably, he contributes to high-performance projects like KeOps, implementing compile-time tensor dot operations in C++ for CPU/GPU autodiff workflows. He also volunteers as technical lead for LAAB, designing plant-driven reactive systems that combine custom sensors, DMX/Zigbee lighting and sound synthesis—an unusual fusion of art and computation. Trained at CentraleSupélec and versed in history and philosophy of science, he excels at translating abstract methods into practical, well-documented software for researchers and artists.
11 years of coding experience
Diplôme d'ingénieur, Ingénierie informatique, Diplôme d'ingénieur, Ingénierie informatique at CentraleSupélecM1, Histoire et philosophie des sciences et de la technologie, M1, Histoire et philosophie des sciences et de la technologie at Université Paris Cité
English, German, French, Chinese, Arabic
Stackoverflow
23reputation
293reached
0answers
2questions
Github Skills (13)
tensorrt10
metaprogramming10
c-language10
tensor10
tensorflow10
cprogramming-language10
operation10
generics6
rpc6
websharper6
arrayaccess6
single-page-application6
type-constraints6
Programming languages (20)
Github contributions (5)
getkeops/keops
Aug 2018 - Nov 2020
KErnel OPerationS, on CPUs and GPUs, with autodiff and without memory overflows
Role in this project:
Backend Developer
Backend DeveloperContributions:28 commits, 19 pushes, 5 branches in 2 years 3 months
Contributions summary:François-david focused on implementing compile-time tensor dot operations within the KeOps framework. Their contributions involved creating and refining C++ code for tensor manipulation, specifically targeting operations on CPUs and GPUs with autodiff capabilities. The work included developing test cases, optimizing the code, and generalizing the implementation to handle various tensor dimensions and contraction schemes. This involved leveraging C++ metaprogramming techniques and the Tao library to achieve compile-time computation and optimization.
autodiffkernel
fradav/fradav.github.io
Mar 2017 - Feb 2025
Contributions:46 pushes, 2 branches, 1 comment in 8 years
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.