Tony W is a research-focused machine learning engineer based in Paris with six years building and productionizing large-scale multimodal LLMs and retrieval systems. At H Company he led distributed training and systems work for MoE vision-language models up to 235B parameters, improving training efficiency and moving prototypes into production with Kubernetes at scale. He co-authored ColPali (ICLR 2025) and shipped its ecosystem—models, benchmarks, and tooling—with millions of downloads and wide adoption in document retrieval. Comfortable across research and engineering stacks, he bridges large-scale training (FSDP, DeepSpeed, Megatron), inference and serving (vLLM, BentoML, TGI), and data pipelines for synthetic augmentation and automated evaluation. Trained at CentraleSupélec and Cambridge, he pairs rigorous academic grounding with hands-on DevOps and MLOps experience. Outside core ML, he has a track record of leading teams and projects—from rowing crew captaincy to organizing conference workshops—demonstrating practical leadership beyond code.
6 years of coding experience
3 years of employment as a software developer
Master of Engineering - MEng, General Engineering, Master of Engineering - MEng, General Engineering at CentraleSupélec
MPhil in Machine Learning and Machine Intelligence, Artificial Intelligence, MPhil in Machine Learning and Machine Intelligence, Artificial Intelligence at University of Cambridge
Preparatory class, Preparatory class at Lycée Jean Baptiste Say - Higher School Preparatory Classes
French, English, chinois (traditionnel), chinois (simplifié), German
Vision Document Retrieval (ViDoRe): Benchmark. Evaluation code for the ColPali paper.
Contributions:20 releases, 23 reviews, 129 PRs in 8 months
colpaliragretrievalsearchvision-language-model
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.