PeopleofSuperintelligence

Researcher

Doina Precup

Doina Precup is an AI researcher whose reinforcement-learning work studies extended actions and reward design. She studied at the University of Cluj-Napoca and the University of Massachusetts Amherst; her work includes planning with options and the option-critic architecture. At McGill and DeepMind, she led the Montreal office and co-founded AI4Good Lab, a machine-learning program for women and gender-diverse people. Her paper on Markov reward received a 2021 NeurIPS Outstanding Paper Award, and she was elected Fellow of the Royal Society of Canada in 2022. As of May 2026, she is Research Director at Google DeepMind and Canada CIFAR AI Chair at McGill University and Mila.

Works
491
Citations
23,449
h-index
52

Publication numbers from OpenAlex, as of 14 October 2026.

Photo of Doina Precup

Timeline

  1. 1995–2000

    Training in Cluj and Amherst

    She earned an MA in Science at the University of Cluj-Napoca in 1995, followed by an MA and a PhD at the University of Massachusetts Amherst in 1997 and 2000.

  2. 1998

    Planning with extended actions

    She coauthored a 1998 NeurIPS paper with Richard Sutton, Satinder Singh, and Balaraman Ravindran that proposed a framework for switching among options and planning with high-level controllers.

  3. Since 2016

    Training new AI learners

    She and Angelique Mannella began planning AI4Good Lab in 2016. Its first group of 25 participants met at McGill for a machine-learning bootcamp for women and gender-diverse people.

  4. 2017

    Learning options from data

    With Pierre-Luc Bacon and Jean Harb, she coauthored The Option-Critic Architecture, which learns each option's internal policy and termination condition alongside a policy over options.

  5. October 2017

    Leading DeepMind in Montreal

    She led DeepMind's Montreal office while continuing to teach and conduct research at McGill.

  6. 2021

    Reward design earns recognition

    With six coauthors, she studied how Markov reward functions can express tasks in the paper On the Expressivity of Markov Reward, which received a 2021 NeurIPS Outstanding Paper Award.

  7. 2022

    Fellowship and recognition

    She has been a Senior Fellow of CIFAR since 2017 and was elected Fellow of the Royal Society of Canada in 2022.

  8. As of May 2026

    Leading research in Montreal

    She is Research Director at Google DeepMind and Canada CIFAR AI Chair at McGill University and Mila.

Sources and credits

Photo: Robincantin, via Wikimedia Commons. License: CC BY-SA 4.0. Source: commons.wikimedia.org. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. Bicentennial Mini-Science: Artificial Intelligence and the Future of Reason | Faculty of Science - McGill University mcgill.ca. Checked 14 October 2026.
  2. Improved Switching among Temporally Abstract Actions papers.neurips.cc. Checked 14 October 2026.
  3. Ten Years of AI4Good Lab: The Co-Founders on What They Built and Why It Still Matters mila.quebec. Checked 14 October 2026.
  4. The Option-Critic Architecture ojs.aaai.org. Checked 14 October 2026.
  5. Q&A with Doina Precup | Newsroom - McGill University mcgill.ca. Checked 14 October 2026.
  6. NeurIPS 2021 Awards neurips.cc. Checked 14 October 2026.
  7. Doina Precup gairdner.org. Checked 14 October 2026.