PeopleofSuperintelligence

Researcher

Andrey Zhmoginov

Andrey Zhmoginov is a researcher whose work spans plasma physics, computer vision, and language models. He earned a physics bachelor's degree in Russia and a Princeton PhD, then studied antihydrogen formation and trapping at Berkeley. At Google Research, he coauthored MobileNetV2 for mobile vision and HyperTransformer for few-shot learning. His 2025 papers studied how new information shapes LLM knowledge and presented MELODI, a memory architecture for long documents; in 2026, he coauthored MIMESIS for interactive-agent training. Since 2024, he is a research scientist at Google DeepMind.

Works
65
Citations
30,487
h-index
18
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2004

    A physics bachelor's degree

    He earns a B.S. in 2004 from Lobachevsky State University of Nizhni Novgorod's Advanced School of General and Applied Physics.

  2. 2011

    A Princeton plasma PhD

    In 2011, he earns a PhD from Princeton University's Department of Astrophysical Sciences under Nathaniel Fisch. He studies alpha-channeling in mirror machines, wave propagation in plasmas, and laser-plasma interaction.

  3. 2012–2015

    Physics postdoc at Berkeley

    From 2012 to 2015, he works as a postdoctoral scholar in UC Berkeley's Physics Department and Professor Wurtele's group, studying antihydrogen formation and trapping through numerical studies.

  4. 2015–2024

    Google Research and MobileNetV2

    From 2015 to 2024, he works at Google Research as a research software engineer and research scientist. In 2018, he coauthors MobileNetV2, a mobile architecture with inverted residuals and linear bottlenecks.

  5. 2022

    HyperTransformer for few-shot learning

    In 2022, he coauthors HyperTransformer, a Transformer-based model that generates convolutional neural network weights from support samples for supervised and semi-supervised few-shot learning.

  6. 2025

    LLM knowledge and MELODI

    In 2025, he coauthors How New Data Permeates LLM Knowledge and How to Dilute It, which studies how new information can prime an LLM to use it in unrelated contexts. He also coauthors MELODI, a hierarchical memory architecture that processes long documents through short context windows.

  7. Since 2024

    Research at Google DeepMind

    Since 2024, he is a research scientist at Google DeepMind. His research interests include LLMs and Transformers, meta-learning, computer vision, and generative models. In 2026, he coauthors MIMESIS, a user simulator trained on human conversations with 13 realistic behavioral patterns for training interactive agents.

Career

  1. Since 2024Google DeepMindResearch Scientist
  2. 2017–2023Google
  3. 2015–2024Google ResearchResearch Software Engineer and Research Scientist
  4. 2012–2015University of California, BerkeleyPostdoctoral Scholar
  5. 2008–2014Office of Science

Education

  1. 2011Princeton UniversityPhD
  2. 2004Lobachevsky State University of Nizhny NovgorodB.S.

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Andrey Zhmoginov. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. 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.
  3. 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.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Andrey Zhmoginov · Curriculum Vitae azhmogin.github.io. Checked 21 October 2026.
  2. Andrey Zhmoginov · Home page azhmogin.github.io. Checked 21 October 2026.
  3. APPM Department Colloquium - Andrey Zhmoginov colorado.edu. Checked 21 October 2026.
  4. MobileNetV2: Inverted Residuals and Linear Bottlenecks openaccess.thecvf.com. Checked 21 October 2026.
  5. HyperTransformer: Model Generation for Supervised and Semi-Supervised Few-Shot Learning research.google. Checked 21 October 2026.
  6. MELODI: Exploring Memory Compression for Long Contexts deepmind.google. Checked 21 October 2026.
  7. Google at ICLR 2025 research.google. Checked 21 October 2026.
  8. How new data permeates LLM knowledge and how to dilute it arxiv.org. Checked 21 October 2026.
  9. MIMESIS: Learning User Simulators as Training Environments for Interactive Agents arxiv.org. Checked 21 October 2026.
  10. MIMESIS: Learning User Simulators as Training Environments for Interactive Agents viethoang1512.github.io. Checked 21 October 2026.