10 Google DeepMind reinforcement learning researchers

For seven of these Google DeepMind researchers, the work is reinforcement learning for Gemini and other language models, from RL for reasoning and agentic behavior to RLHF and fine-tuning on execution feedback. The other three study exploration and fast adaptation, world models and generative modeling, or sat on DeepMind's Reinforcement Learning Team. They fit AI teams deciding how to post-train a model with RL, build RL infrastructure or evaluate agents.

10 professionals1 companyData retrieved October 8, 2026

Professionals to explore

  1. Abhinav Gupta

    LinkedIn

    Experience: Research Scientist · Google DeepMind

    $808 per 30-minute call

    Part of the core Gemini team, Abhinav Gupta has been a Research Scientist at Google DeepMind since 2024 and works at the intersection of reinforcement learning and language, fine-tuning LLMs with machine and execution feedback and building robust evaluation metrics. Abhinav was a Doctoral Student at Mila - Quebec Artificial Intelligence Institute (2018-2024).

    Is this you? Claim your profile and get paid for calls.

  2. Abram L. Friesen

    LinkedIn

    Experience: Senior Staff Research Scientist · Google DeepMind

    $2,019 per 30-minute call

    Abram L. Friesen works on Gemini at Google DeepMind in London, focused on reinforcement learning for improving reasoning, agentic behavior and overall model capabilities, and describes a key role in training Gemini 3 and improving reasoning for Gemini 2.5 Pro and Flash. Abram has been a Staff Research Scientist since 2019 and a Senior Staff Research Scientist since 2026.

    Is this you? Claim your profile and get paid for calls.

  3. Aditi Mavalankar

    LinkedIn

    Experience: Research Scientist · Google DeepMind

    $808 per 30-minute call

    Aditi Mavalankar has been a Research Scientist at Google DeepMind since 2022, after two Research Scientist Intern stints at DeepMind (2021 and 2022). Aditi's research is on how agents extract useful information from interactions with the environment for faster adaptation, generalization or exploration, the pick for exploration questions.

    Is this you? Claim your profile and get paid for calls.

  4. Avi Singh

    LinkedIn

    Experience: Research Scientist · Google DeepMind

    $808 per 30-minute call

    Avi Singh is interested in reinforcement learning for large language models and has been a Research Scientist at Google DeepMind since 2023, after Research Scientist at Google (2021-2023). Avi was a Graduate Student Researcher at the University of California, Berkeley (2016-2021), advised by Sergey Levine.

    Is this you? Claim your profile and get paid for calls.

  5. Bobak Shahriari

    LinkedIn

    Experience: Staff Research Scientist · Google DeepMind

    $1,346 per 30-minute call

    Online and offline RL fine-tuning of LLMs and multimodal LLMs, including RLHF and RLAIF, is Bobak Shahriari's recent work as a Research Scientist at Google DeepMind since 2018 and Staff Research Scientist since 2023. Bobak also released Acme, a framework for RL research with a large repository of online and offline RL agents, which suits RL infrastructure questions.

    Is this you? Claim your profile and get paid for calls.

  6. Matthew W. Hoffman

    LinkedIn

    Experience: Senior Staff Research Scientist · Google DeepMind

    $2,019 per 30-minute call

    Senior Staff Research Scientist at Google DeepMind since 2016, Matthew W. Hoffman works on systems for training large-scale LLMs with online and offline RL. Earlier work focused on research-oriented systems for RL, particularly in continuous-action domains and general RL algorithms, and Matthew held a Postdoctoral Researcher role at the University of Cambridge (2013-2016).

    Is this you? Claim your profile and get paid for calls.

  7. Samuel Holt

    LinkedIn

    Experience: Research Scientist · Google DeepMind

    $808 per 30-minute call

    Samuel Holt has been a Research Scientist at Google DeepMind since 2025, working on LLM agents and their learning dynamics with a particular focus on reinforcement learning, toward universal AI assistants that remember context and use tools. Samuel was a PhD Candidate in Machine Learning at the University of Cambridge (2021-2025) and a Google DeepMind student researcher (2024).

    Is this you? Claim your profile and get paid for calls.

  8. Sherry Yang

    LinkedIn

    Experience: Staff Research Scientist · Google DeepMind

    $1,346 per 30-minute call

    Staff Research Scientist at Google DeepMind since 2023, listing reinforcement learning, generative modeling, world models and agents as research areas. Sherry Yang was a Research Scientist at Google (2018-2023), including Google Brain, so world models and model-based agent questions fit.

    Is this you? Claim your profile and get paid for calls.

  9. Yuchen Zhuang

    LinkedIn

    Experience: Research Scientist · Google DeepMind

    $808 per 30-minute call

    RL and post-training are the focus Yuchen Zhuang names as a Research Scientist at Google DeepMind since 2025, aiming at language model agents capable of human-like reasoning and planning. Yuchen was an Applied Scientist at Amazon (2024) and a Research Scientist Intern at Adobe (2023), a fit for questions on agent post-training.

    Is this you? Claim your profile and get paid for calls.

  10. Zita Marinho

    LinkedIn

    Experience: Former Research Scientist · Google DeepMind

    $808 per 30-minute call

    Former Research Scientist at Google DeepMind (2020-2026) on the Reinforcement Learning Team, Zita Marinho has been an Assistant Professor at Universidade de Lisboa (University of Lisbon) since 2026. Zita was a Senior Research Scientist at Priberam (2018-2020) and a PhD Student at Carnegie Mellon University (2012-2018), useful for comparing RL research inside a lab and in academia.

    Is this you? Claim your profile and get paid for calls.

Choose the right perspective

For RL post-training of large models at scale, Abram L. Friesen, Matthew W. Hoffman and Bobak Shahriari have the longest records at the lab. Yuchen Zhuang, Avi Singh, Abhinav Gupta and Samuel Holt work on RL for language models and agents, while Aditi Mavalankar, Sherry Yang and Zita Marinho are better for classic RL questions on exploration, world models and generalization.

Questions to take into the conversation

  1. 01When post-training a large language model, how do Google DeepMind researchers choose between online RL, offline RL and preference-based methods like RLHF?
  2. 02What reward signals work best for improving reasoning and agentic behavior, and how do you catch reward hacking early?
  3. 03How should a team building RL agents on top of language models set up evaluation so gains hold outside the training tasks?

About this directory

This is a professional research starting point based on business profile data retrieved on . Titles and companies reflect that source snapshot; roles marked former had ended by then. Check the linked profiles for current details. Inclusion does not imply Instant Expert membership or availability.

Other perspectives in Artificial intelligence