Irina Rish  is a Full Professor of Computer Science at Université de Montréal (UdeM), director of its Autonomous AI Lab, and core faculty member of MILA - Quebec AI Institute. She holds the Canada Excellence Research Chair (CERC) in Autonomous AI and a Canada CIFAR AI Chair, an MSc and PhD in AI from the University of California, Irvine, and an MSc in Applied Mathematics from Moscow Gubkin Institute.

Her research spans automated reasoning, probabilistic inference in graphical models, machine learning, sparse modelling, and neuroscience-inspired AI. Her current work advances reliable, continually learning AI through research on robustness, interpretability, model compression, and how foundation-model capabilities and behaviors evolve with scale across language, vision, and time series. Recent lab research includes forecasting with numerical data and textual context, adaptation to changing data without forgetting, defenses against malicious instructions encountered by AI agents, and collusion risks among autonomous agents making market decisions, addressing opportunities for better prediction and efficiency alongside challenges in model evaluation and oversight.

She has received major U.S. Department of Energy compute awards, including the 2023 INCITE award and multiple ALCC awards, and led open-source foundation-model projects on Summit and Frontier at Oak Ridge Leadership Computing Facility, including the 2023 Scalable Foundation Models INCITE project. She advises 42.com on multimodal foundation models for finance, Nolano.ai on efficient time-series foundation models, and MUUTAA on AI for healthcare supply-chain optimization.

Before joining UdeM in 2019, she spent two decades as a research scientist at IBM’s T.J. Watson Research Center, working at the intersection of neuroscience and AI and leading the Neuro-AI challenge. Her IBM honors include the Eminence & Excellence and Outstanding Innovation Awards (2018), Outstanding Technical Achievement Award (2017), and Research Accomplishment Award (2009). She holds 64 patents, has authored over 170 research papers and contributed book chapters, edited three books, and published a monograph on Sparse Modeling.