PeopleofSuperintelligence

Researcher

Anna Harutyunyan

Anna Harutyunyan is a reinforcement learning researcher whose work examines off-policy learning, credit assignment, and the limits of reward functions. Born and raised in Armenia, she turned to reinforcement learning in 2013 and completed a PhD at the Free University of Brussels in 2018. She coauthored Retrace and Hindsight Credit Assignment, and her 2021 Markov reward paper earned a NeurIPS Outstanding Paper Award. Her later research analyzed quantile learning, plasticity, and coherence, and she is a research scientist at Google DeepMind as of October 2026.

Badges

Timeline

  1. 2013

    Turning to reinforcement learning

    Born and raised in Armenia, she turns to reinforcement learning in 2013.

  2. 2016

    Retrace for off-policy learning

    With Rémi Munos, Tom Stepleton, and Marc G. Bellemare, she coauthors Retrace in 2016, an off-policy algorithm with low variance, safe use of behavior-policy samples, and efficient use of near on-policy data.

  3. 2018

    A PhD in Brussels

    In 2018, she earns a PhD from the Free University of Brussels, with a thesis titled “Beyond single step temporal difference learning.”

  4. 2019

    Hindsight credit assignment

    In 2019, she coauthors Hindsight Credit Assignment, which assigns credit to past decisions based on their likelihood of leading to an observed outcome.

  5. 2021

    A NeurIPS Outstanding Paper

    In 2021, she coauthors On the Expressivity of Markov Reward, which shows that some tasks have no Markov reward function that can capture them. The paper earns a NeurIPS Outstanding Paper Award.

  6. 2024

    Quantile temporal-difference theory

    In 2024, she coauthors a theoretical analysis of quantile temporal-difference learning, a distributional reinforcement-learning algorithm. It proves convergence with probability 1 to fixed points of related dynamic-programming procedures.

  7. 2025

    An anti-tutorial and plasticity

    At Yerevan State University’s 2025 summer school, she gives “Reinforcement learning: an anti-tutorial” as an invited lecturer from Google DeepMind. Her NeurIPS 2025 paper Plasticity as the Mirror of Empowerment defines plasticity as an agent’s capacity to be influenced by observations and connects it with empowerment.

  8. As of October 2026

    Research scientist at Google DeepMind

    In 2026, she coauthors Cohering Reinforcement Learning, which proposes coherence as a principle for agents in non-Markov worlds and explores complex-valued rewards. As of October 2026, she is a research scientist at Google DeepMind; a June 2026 joint statement on AI in war lists her affiliation as Google DeepMind.

Career

  1. NowGoogle DeepMindResearch scientist
  2. 2025Yerevan State UniversityInvited lecturer

Education

  1. 2018Vrije Universiteit BrusselPhD

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Anna Harutyunyan. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Anna Harutyunyan. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Anna Harutyunyan yet. Send us a free-licensed photo.

  1. Speakers imol-workshop.github.io. Checked 21 October 2026.
  2. Safe and Efficient Off-Policy Reinforcement Learning arxiv.org. Checked 21 October 2026.
  3. Hindsight Credit Assignment arxiv.org. Checked 21 October 2026.
  4. On the Expressivity of Markov Reward — Google DeepMind deepmind.google. Checked 21 October 2026.
  5. Brown CS Alums David Abel and Mark K. Ho, Professor Michael Littman, and Collaborators Win a NeurIPS Outstanding Paper Award awards.cs.brown.edu. Checked 21 October 2026.
  6. An Analysis of Quantile Temporal-Difference Learning jmlr.org. Checked 21 October 2026.
  7. “Mathematics and Applications”: 7th summer school ysu.am. Checked 21 October 2026.
  8. Plasticity as the Mirror of Empowerment papers.nips.cc. Checked 21 October 2026.
  9. Cohering Reinforcement Learning rlj.cs.umass.edu. Checked 21 October 2026.
  10. 戦争におけるAIに関する共同声明 jca.apc.org. Checked 21 October 2026.