Holger Schwenk
Reserach Director At Meta AI
Greater Paris Metropolitan Region France
Summary
🤩
RockstarHolger Schwenk is a research director at Meta AI with over a decade of experience bridging academic research and industrial ML productization in speech recognition and machine translation. He previously served as a research scientist at Facebook and co-founded DeepLingo, where he built customized MT engines for clients, and earlier led a university computer science lab as professor. Holger is a hands-on practitioner who contributes to open-source projects like Facebook Research's LASER, updating PyTorch compatibility and training pipelines that keep large multilingual models operational. His background combines deep neural NLP expertise with practical engineering—moving models from prototype to robust pipeline deployments. Based in the Greater Paris area, he brings both leadership in large research organizations and founder-level engineering agility. An understated detail: his work often focuses on language-agnostic sentence representations that enable cross-lingual transfer at scale.
7 years of coding experience
14 years of employment as a software developer
English, French, German
Github Skills (6)
pytorch10
machine-learning10
nlp10
python9
bash9
script9
Programming languages (4)
Github contributions (5)
facebookresearch/LASER
Jun 2018 - Jul 2022
Language-Agnostic SEntence Representations
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 76 commits, 14 PRs in 4 years 1 month
Contributions summary:Holger made several updates related to PyTorch compatibility, indicating involvement in the project's machine-learning aspects. They updated scripts related to sentence classification, incorporating changes for PyTorch 0.40 and PyTorch 1.0. Additionally, they updated various shell scripts related to data processing and the XNLI task, suggesting they are also involved in model training and evaluation pipelines. The contributions demonstrate a focus on maintaining and improving the core machine learning infrastructure of the project.
nlpsentencerepresentationsagnosticsentence-embeddings
facebookresearch/MLDoc
Jun 2018 - Feb 2019
A Corpus for Multilingual Document Classification in Eight Languages.
Contributions:1 comment in 8 months
nlpcorpusdocument-classificationsimilaritysentence-similarity
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.