Lucile Saulnier
AI Specialist at Mistral AI
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolLucile Saulnier is an AI specialist with six years of experience building and researching large multilingual and multimodal models, currently contributing at Mistral AI after impactful engineering work at Hugging Face. She played a hands-on role in developing BLOOM and the ROOTS dataset during the BigScience workshop and worked on tokenization improvements in the widely used Hugging Face transformers library. Trained at CentraleSupélec and ENS Paris-Saclay (MVA), she combines strong mathematical foundations with applied ML engineering to move research into production. Her background spans privacy-aware healthcare ML at Arkhn to venture and teaching roles, reflecting an unusual blend of technical depth, product-mindedness, and communication skills. Notably, her tokenizer work addressed tricky multilingual and BPE edge cases (including Chinese handling), boosting robustness in one of the ecosystem’s cornerstone repos.
6 years of coding experience
3 years of employment as a software developer
Centrale Paris; Master's degree, Science and Executive Engineering, Centrale Paris; Master's degree, Science and Executive Engineering at CentraleSupélecMaster 2 (M2), Mathématiques, Vision, Apprentissage (MVA), mathématiques appliquées, Master 2 (M2), Mathématiques, Vision, Apprentissage (MVA), mathématiques appliquées at École normale supérieure Paris-SaclayClasse préparatoirs aux grandes écoles, Mathematics and Physics, Classe préparatoirs aux grandes écoles, Mathematics and Physics at Lycée Charlemagne (75004)Classe préparatoire aux grandes écoles, Mathematics and Physics, Classe préparatoire aux grandes écoles, Mathematics and Physics at Lycée Condorcet (75009)
English, Spanish, French
Github Skills (14)
transformers10
tokenize10
pytorch10
machine-learning10
tokenizer10
nlp10
language-model10
tensorflow10
python10
natural-language-processing10
bert10
clip9
gpt9
roberta9
Programming languages (7)
Github contributions (5)
huggingface/transformers
Jun 2021 - Nov 2022
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:
ML Engineer
ML EngineerContributions:233 reviews, 30 commits, 71 PRs in 1 year 5 months
Contributions summary:Lucile primarily contributed to the `transformers` repository, focused on enhancing the tokenization functionalities of various models. Their work involved implementing new features related to the `PreTrainedTokenizerFast` class, adapting common tests, and improving the handling of special tokens and configuration files. They also addressed issues related to tokenization in Byte-Pair-Encoding (BPE) models and ensured correct behavior when handling Chinese characters, and made other enhancements and bug fixes related to tokenizers.
pythonbertspeech-recognitionstate-of-the-artflax
SaulLu/bigscience
Sep 2021 - Feb 2022
Codebase for the engineering/scaling WG
Contributions:14 commits, 20 pushes, 9 branches in 5 months
codebasepythonengineeringscaling
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.