Pascal Poupart
at University of Waterloo, Vector Institute
Professor at University of Waterloo & Canada CIFAR AI Chair at Vector Institute; researcher in reinforcement learning, machine learning, and NLP
Canada Themes
Pascal Poupart is a Professor in the David R. Cheriton School of Computer Science at the University of Waterloo, where he conducts research in reinforcement learning, machine learning, natural language processing, and artificial intelligence. He also serves as Research Director and Canada CIFAR AI Chair at the Vector Institute, a Canadian research organization focused on machine learning and AI.
His work spans both theoretical and applied aspects of AI, with a particular emphasis on probabilistic methods and decision-making under uncertainty. Poupart is affiliated with the Waterloo AI Institute and can be contacted through the University of Waterloo, where his research group and publications are documented on his academic profile.
Academic Background and Career Path
Poupart's academic formation took him across three of Canada's leading research universities. He earned his undergraduate degree in computer science from McGill University in 1998, followed by a Master's degree from the University of British Columbia in 2000, and completed his doctoral work at the University of Toronto in 2004. This trajectory through institutions with strong foundations in artificial intelligence and machine learning shaped his research orientation toward probabilistic reasoning and decision-making frameworks that continue to define his scholarly identity.
Beyond his academic appointment at Waterloo, Poupart spent two years (2018–2020) as Research Director and Principal Research Scientist at the Waterloo Borealis AI Research Lab, a facility funded by the Royal Bank of Canada. This engagement with an industry-backed research environment reflects a recurring thread in his career: maintaining active connections between fundamental research and practical application across sectors including finance, healthcare, and technology.
Research Projects and Focal Areas
Within the broad landscape of machine learning, Poupart's group concentrates on several interrelated threads. On the methodological side, active projects include probabilistic deep learning, robust machine learning, and data-efficient reinforcement learning — work aimed at making learning systems more reliable and sample-efficient in real-world conditions. His team also investigates federated learning, continual learning, self-supervised learning, meta-learning, few-shot learning, causal learning, and uncertainty quantification, reflecting engagement with many of the open problems in contemporary machine learning research.
Applied threads are equally prominent. In natural language processing, the group works on conversational agents, automated document editing, and grammar error correction. A more unexpected application domain is material design, where Poupart's team applies Bayesian optimization to identify promising catalysts, oxygen carriers, and other materials that facilitate chemical reactions relevant to CO₂ conversion and CO₂ capture. Sports analytics, specifically work with data from hockey, represents another applied direction. The breadth of these applications illustrates how probabilistic machine learning methods transfer across domains ranging from climate-related chemistry to professional sports performance analysis.
Industry Collaboration and Advisory Roles
Poupart has maintained an unusually wide network of research collaborators and advisory relationships spanning both technology and regulated industries. He has served as a scientific advisor to companies including ElementAI, DialPad, ProNavigator, and Scribendi. Collaborative research partnerships have extended to organizations such as Google, Microsoft, Intel, Ford, Manulife, the Royal Bank of Canada, and the Bank of Montreal, as well as conversational technology companies Kik Interactive and In the Chat.
On the healthcare side, his collaborators have included the Alzheimer Association, the UW-Schlegel Research Institute for Aging, Sunnybrook Health Science Centre, and the Toronto Rehabilitation Institute — connections that align with his earlier research interest in health informatics. This combination of technology-sector and healthcare partnerships positions his applied work at intersections where uncertainty quantification and adaptive decision-making carry direct consequences.
Recognition and Teaching
Poupart received the Ontario Early Researcher Award, held from 2008 to 2013, and was the recipient of the David R. Cheriton Faculty Fellowship between 2015 and 2018. Competitive recognition includes a best paper award runner-up at UAI in 2008, and success in the SAT 2016 Competition, where his team claimed both the best main track solver and best application solver distinctions, with a runner-up best student paper award following at SAT 2017.
His teaching record at Waterloo is extensive, spanning courses in introductory artificial intelligence, machine learning, and more specialized graduate topics. Courses such as CS486 (Introduction to Artificial Intelligence) and CS885 (Reinforcement Learning) appear repeatedly across more than fifteen years of teaching records, indicating sustained commitment to educating students at both undergraduate and graduate levels in the foundations and frontiers of AI.