Summary
Samuel Lavoie is a research scientist based in Montreal with a decade of experience developing large probabilistic and discrete representation models for multimodal and generative tasks. His work spans influential papers (Llip, SEM, DLC) on multimodal encoder objectives and discrete latent codes for image generation, and has bridged academic and industry labs through roles at Mila, Meta, and now Google. Trained in software engineering and computer science with a PhD track at Mila, he combines rigorous theoretical grounding with practical engineering from internships at Microsoft and industry experience. Notably, he has focused on discrete representation learning—an area that enables efficient, composable generation beyond standard continuous latent approaches—highlighting a practical bent toward scalable, modular generative systems.
10 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy - PhD, Doctor of Philosophy - PhD at Mila - Quebec Artificial Intelligence Institute
Master's degree, Computer Science, Master's degree, Computer Science at Université de Montréal
Bachelor of Engineering (B.Eng.), Software Engineering, Bachelor of Engineering (B.Eng.), Software Engineering at Polytechnique Montréal
Health science, Science, Health science, Science at College de Maisonneuve
French, English