Thomas Chaton
Senior Staff Research Engineer (IC7) at Lightning AI
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolThomas Chaton is a Senior Staff Research Engineer with a decade of experience building and scaling deep learning systems, currently advancing deep geometric learning on graphs and manifolds at HELIX RE within Lightning AI. He blends hands-on contributions to major open-source PyTorch projects—helping stabilize PyTorch Lightning, TorchMetrics, and point-cloud tooling—with leadership that grew PyTorch Lightning’s core team and usage dramatically. Equally comfortable in research and production, he has shipped KPConv and deformable point-cloud modules, fixed critical training and checkpointing bugs for multi-device workflows, and improved metric synchronization for distributed training. His background spans cloud-native ML stacks (GCP/AWS, Kubernetes, KubeFlow), model explainability and compression research, and patented work presented internationally, reflecting a rare mix of theoretical depth and pragmatic engineering.
10 years of coding experience
7 years of employment as a software developer
Classe prépartoire aux grandes écoles PTSI-PT* Mathématiques, Classe prépartoire aux grandes écoles PTSI-PT* Mathématiques at Lycée Lamartinière MontplaisirSystem Security/ Data Science GPA 3.87 Machine Learning/Deep Learning, System Security/ Data Science GPA 3.87 Machine Learning/Deep Learning at Télécom ParisData Science / Machine Learning / AI, Data Science / Machine Learning / AI at EURECOMBaccalauréat S Physique, Baccalauréat S Physique at Lycée Jean-Henri Fabre
French, English, Spanish
Stackoverflow
1reputation
0reached
3answers
0questions
Github Skills (27)
pytorch10
checkpoint10
python10
machine-learning10
pytorch-lightning10
model-checking10
checkpointing10
data-preprocessing10
deep-learning10
point-cloud10
metric10
documentation10
classification9
deeplearning-ai9
computer-vision9
Programming languages (17)
Github contributions (5)
torch-points3d/torch-points3d
Dec 2019 - Mar 2021
Pytorch framework for doing deep learning on point clouds.
Role in this project:
ML Engineer
ML EngineerContributions:5 reviews, 659 commits, 252 PRs in 1 year 3 months
Contributions summary:Thomas implemented the PointKernel module, a core component for the KPConv architecture, and integrated it within the larger framework for deep learning on point clouds. They contributed to the implementation of the deformable KPConv by integrating offsets and various loss functions and replaced existing einsum equations. Furthermore, they addressed multiple bugs related to the architecture's inner workings, showing a dedication to debugging and optimization.
pytorchscannetpytorch-frameworkpoint-cloudss3dis
Lightning-Universe/lightning-flash
Jan 2021 - Dec 2021
Your PyTorch AI Factory - Flash enables you to easily configure and run complex AI recipes for over 15 tasks across 7 data domains
Role in this project:ML Engineer & Data Scientist
ML Engineer & Data ScientistContributions:1 release, 903 reviews, 111 commits in 11 months
Contributions summary:Thomas's commits primarily involved modifying and improving the `pl_flash/tabular/data/data.py`, `pl_flash/vision/data/classification.py`, and test files, indicating involvement in data loading, processing, and dataset creation within the PyTorch Lightning Flash framework. They also added examples for torchvision classifier. The changes focused on improving the tabular data functionality, integrating video classification features and the addition of predict functionality for various tasks like image and text classification.
torch-geometricicevisionconfigurepytorch-lightningdomains
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.