Simon Caby
Ingénieur at Self-employed
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolSimon Caby is an engineer and biology-inspired neural models enthusiast based in Paris, blending deep expertise in spiking and artificial neural networks with decades of low-level programming experience (Python, C/C++, CPU/GPU assembly) honed in the video game industry. He has led technical teams and founded companies across gaming, interactive media, and music production, scoring award-winning film and TV projects including work for Studio Ghibli and multiple gold/platinum records. As a contributor to BindsNET he optimized PyTorch-based SNN forward passes and resolved CPU/GPU placement issues, reflecting a focus on performant, research-grade ML code. Comfortable switching between research, production engineering, and creative direction, he uniquely bridges neuroscience-inspired AI, graphics, and audio composition.
7 years of coding experience
27 years of employment as a software developer
Ecole AlsacienneMaster's degree, Mathematics and Computer Science, Master's degree, Mathematics and Computer Science at Université Denis Diderot (Paris VII)Conservatoire de musique de Paris XIV
English, French, German, Chinese
Stackoverflow
191reputation
4kreached
15answers
0questions
Github Skills (17)
pytorch10
gpgpu10
neurons10
machine-learning10
parallel-computing10
gpu10
accelerated-computing10
spiking-neural-networks10
simulation9
python9
simulations9
neural-network6
keras6
tensorflow6
loss-functions6
Programming languages (4)
Github contributions (5)
BindsNET/bindsnet
May 2020 - Dec 2020
Simulation of spiking neural networks (SNNs) using PyTorch.
Role in this project:
ML Engineer
ML EngineerContributions:15 commits, 13 PRs, 40 pushes in 7 months
Contributions summary:Simon focused on optimizing code related to spiking neural networks (SNNs) within the PyTorch framework. Their contributions involved performance improvements, specifically in the forward pass of LIF nodes, Diehl & Cook, SRM0 nodes, and CurrentLIF & IF nodes. They refactored code, corrected a default reduction, and addressed CPU/GPU placement issues. These changes aimed to enhance the efficiency and performance of the SNN simulations.
pytorchspiking-neural-networksdynamicreinforcement-learningstdp
SimonInParis/SNN_WTA
Jun 2020 - Jun 2020
A modified WTA learning scheme with BindsNet
Contributions:4 PRs, 12 pushes, 2 branches in 1 day
schemewtamachine-learningmodified
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.