Hugo Touvron

Paris, Ile-de-France

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Hugo Touvron is a research scientist at Meta AI with six years of experience building and extending cutting-edge computer vision models. He completed a PhD at Facebook AI Research and Sorbonne University and holds engineering degrees from École Polytechnique and ENSTA, plus a Master 2 MVA from ENS Paris-Saclay. Hugo has contributed to high-profile open-source projects such as DeiT—adding architectures like CaiT and ResMLP—and adapted EfficientNet integrations in the popular FixRes repository, demonstrating both model design and training-pipeline expertise. His work blends rigorous academic research (see his Google Scholar) with hands-on engineering that pushes state-of-the-art image classification and fine-tuning practices. Based in Paris, he combines a strong theoretical foundation with practical coding contributions that help move research models toward reproducible, production-ready implementations.

6 years of coding experience

Sorbonne University

French, English, Spanish

Github Skills (17)

transformers10

pytorch10

python10

model-driven10

image-classification10

machine-learning10

model-building10

deep-learning10

vision-transformer10

computer-vision10

transformer10

efficientnet10

model-driven-development10

distributed-training9

modeling9

Programming languages (1)

Python

Github contributions (5)

facebookresearch/FixRes

Aug 2019 - Dec 2020

This repository reproduces the results of the paper: "Fixing the train-test resolution discrepancy" https://arxiv.org/abs/1906.06423

Role in this project:

ML Engineer

Contributions:24 commits, 3 PRs, 24 pushes in 1 year 5 months

Contributions summary:Hugo made several modifications to the `imnet_finetune/train.py` file, which included integrating and configuring EfficientNet models within a training pipeline. They also adjusted the `main_finetune.py` and `config.py` files, likely to support the EfficientNet integration by adding new parameters and configurations. Furthermore, the user updated the `inception.py` module. This suggests a focus on modifying and adapting existing model architectures and training infrastructure, possibly for a specific image classification or fine-tuning task related to the repository's objective.

pytorcharxivloss-functionsabsefficientnet

facebookresearch/deit

Dec 2020 - Oct 2022

Official DeiT repository

Role in this project:

ML Engineer

Contributions:9 reviews, 70 commits, 19 PRs in 1 year 10 months

Contributions summary:Hugo primarily contributed to the development and enhancement of computer vision models within the repository. Their work involved adding new model architectures such as CaiT and ResMLP, along with implementing corresponding model definitions and pre-trained weights. They also addressed existing model implementations and integrated new features into the codebase, demonstrating a focus on expanding the model library and ensuring the functionality of existing components.

deit

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.

Request Free Trial