Machine learning · Forecasting · Enterprise AI
I build learning machines, then the companies around them.
Co-founder and Chief Scientist at Nera Software. Adjunct Professor at Polytechnique Montréal and Associate Industry Member of Mila. Previously VP Research at ServiceNow. Co-founder of Element AI, Imagia, ApSTAT and Chapados Couture Capital.
Selected work
N-BEATSInterpretable deep forecasting; the most cited of these works. · International Conference on Learning Representations 2020
StarCoder 2 and The Stack v2Open code models and their training data, with the BigCode community. · BigCode Project Technical Report 2024
LLM2VecTurning decoder-only language models into strong text encoders. · First Conference on Language Modeling (COLM 2024) 2024
WorkArenaBenchmarking web agents on real knowledge-work tasks. · Proceedings of the 41st International Conference on Machine Learning (ICML 2024) 2024
Speaking
Context is Key: A Benchmark for Forecasting with Essential Textual InformationIJCAI 2025 Workshop on Artificial Intelligence for Time Series Analysis (AI4TS), Montréal, Canada · August 2025
Agentic Full-Stack Benchmarking for Knowledge WorkACL 2025 Workshop for Research on Agent Language Models (REALM), Vienna, Austria · July 2025
AI Agents for Enterprise WorkflowsUC Berkeley, CS294/194-196 Large Language Model Agents (course and MOOC) · October 2024
News