Benjamin Thérien
Research Scientist Intern at Mila - Quebec Artificial Intelligence Institute
Montreal, Quebec, Canada
Summary
🤩
Rockstar🎓
Top SchoolBenjamin Thérien is a PhD student at Université de Montréal and Mila, currently interning as a Research Scientist at Meta where he focuses on distributed optimization for cross-datacenter training of large language models. With seven years of industry and research experience spanning Meta, Capital One, Morgan Stanley, and multiple Mila internships, he brings a rare blend of systems engineering and scalable ML research—especially in LLM conversion, MoE pre-training, and high‑performance parallelism. His prior work includes converting transformers to GPT‑NeoX formats, optimizing 3D parallelism on supercomputers for 9B-parameter models, and creating a paired LiDAR-image dataset that underpinned a WACV publication. Comfortable shipping production tooling and large-scale experiments, he pairs strong software design instincts (API/automation and data pipelines) with rigorous academic output. A US and Canadian citizen based in Montreal, he’s as likely to optimize GPU TFLOPs on Summit as to lead end-to-end research projects from idea to publication.
7 years of coding experience
3 years of employment as a software developer
MMath, Computer Science, MMath, Computer Science at University of WaterlooCollege Degree, Business/Commerce, General, College Degree, Business/Commerce, General at Marianopolis CollegeBachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at Concordia UniversityDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Universite de Montreal
English, French
Github Skills (73)
transformers10
parallel10
camera10
sensor-fusion10
gpt10
sensor10
lidar10
pytorch10
huggingface-transformers10
deepspeed10
nlp9
python9
language-model9
fairness-ml9
gpu9
Programming languages (4)
Github contributions (5)
bentherien/gpt-neox
Apr 2023 - Apr 2024
An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
Contributions:13 reviews, 18 PRs, 46 pushes in 11 months
bentherien/parametricScatteringNetworks
Jul 2021 - Mar 2022
Contributions:4 commits, 6 pushes, 1 branch in 8 months
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.