PeopleofSuperintelligence

Researcher

Po-Sen Huang

Po-Sen Huang is a research scientist at Google DeepMind whose papers cover web search, speech separation, language models and algorithm discovery. He studied at National Taiwan University and earned a PhD at the University of Illinois at Urbana-Champaign under Mark Hasegawa-Johnson. His 2013 paper on Deep Structured Semantic Models for web search received the CIKM 2024 Test of Time Award, and a 2015 source-separation paper he coauthored received the 2020 IEEE SPS Best Paper Award. After research roles at Clarifai and Microsoft, he joined Google in 2018 and coauthored the Gopher language-model analysis, the AlphaCode paper and the SynthID-Text watermarking scheme published in Nature in 2024. In 2025, he was among the developers of AlphaEvolve, Google DeepMind's coding agent for algorithm discovery.

Works
77
Citations
7,834
h-index
28
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2009–2015

    PhD student at Illinois

    He studies at the University of Illinois at Urbana-Champaign from 2009 to 2015 and earns a PhD there, advised by Mark Hasegawa-Johnson. He arrives with a B.S. degree from National Taiwan University.

  2. 2013–2024

    Deep structured semantic models

    He coauthors 'Learning Deep Structured Semantic Models for Web Search using Clickthrough Data' with Xiaodong He, Jianfeng Gao, Li Deng, Alex Acero and Larry Heck, presented at CIKM in October 2013. The models project queries and documents into a common low-dimensional space where relevance is computed as a distance. The paper wins the CIKM 2024 Test of Time Award.

  3. 2015–2020

    Monaural source separation

    In 2015, he coauthors a paper with Minje Kim, Mark Hasegawa-Johnson and Paris Smaragdis on joint optimization of masking functions and deep recurrent neural networks for speech separation, singing-voice separation and speech denoising from a single channel. The paper appears in IEEE/ACM Transactions on Audio, Speech, and Language Processing in December 2015 and receives the 2020 IEEE SPS Best Paper Award.

  4. 2015–2018

    From Clarifai to Google

    He works as a research scientist at Clarifai from 2015 to 2016 and as a researcher at Microsoft from 2016 to 2018. In 2018, he joins Google as a research scientist.

  5. 2021–2022

    Gopher, AlphaCode and model risks

    In December 2021, he is among the authors of 'Scaling Language Models: Methods, Analysis & Insights from Training Gopher', an analysis of Transformer language models up to a 280 billion parameter model called Gopher, and of 'Ethical and social risks of harm from Language Models'. In February 2022, he is among the authors of 'Competition-Level Code Generation with AlphaCode', which introduces a system that creates novel solutions to competitive programming problems.

  6. 2024

    SynthID-Text in Nature

    In October 2024, Nature publishes 'Scalable watermarking for identifying large language model outputs', which describes SynthID-Text, a production-ready text watermarking scheme; he is one of its authors at Google DeepMind in London. In November 2024, he is among the authors of a Nature Medicine paper that builds Flamingo-CXR, a report generation system for chest radiographs.

  7. 2025

    The AlphaEvolve coding agent

    In May 2025, Google DeepMind announces AlphaEvolve, an evolutionary coding agent powered by large language models for general-purpose algorithm discovery and optimization, and credits him among its developers. The agent has helped design faster matrix multiplication algorithms and find new solutions to open mathematical problems, and its white paper reaches arXiv in June 2025.

  8. As of October 2026

    Research scientist at DeepMind

    He is a research scientist at Google DeepMind, the lab that announced AlphaEvolve in 2025.

Career

  1. Since 2019Google DeepMindResearch Scientist
  2. 2018–2024GoogleResearch Scientist
  3. 2016–2018MicrosoftResearcher
  4. 2015–2016ClarifaiResearch Scientist
  5. 2012–2015National University of Kaohsiung

Education

  1. 2015University of Illinois Urbana-ChampaignPhD
  2. National Taiwan UniversityB.S.

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

Most-cited works

  1. Learning deep structured semantic models for web search using clickthrough data2013 · ACM International Conference on Information & Knowledge Management · 2,050 citationsWith Alex Acero and Li Deng
  2. Competition-level code generation with AlphaCode2022 · Science · 1,034 citationsWith Cyprien de Masson d’Autume, Julian Schrittwieser, Koray Kavukcuoglu, Nando de Freitas, Nate Kushman and Oriol Vinyals
  3. Taxonomy of Risks posed by Language Models2022 · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · 744 citationsWith Abeba Birhane, Geoffrey Irving, Iason Gabriel and Lisa Anne Hendricks
  4. Ethical and social risks of harm from Language Models2021 · 476 citationsWith Abeba Birhane, Geoffrey Irving, Iason Gabriel and Lisa Anne Hendricks
  5. Joint Optimization of Masks and Deep Recurrent Neural Networks for Monaural Source Separation2015 · IEEE/ACM Transactions on Audio Speech and Language Processing · 461 citations

From OpenAlex, as of 21 October 2026.

  1. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Po-Sen Huang. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  2. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Po-Sen Huang. 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.
  3. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Po-Sen Huang. 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.
  4. 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.

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 Po-Sen Huang yet. Send us a free-licensed photo.

  1. Po-Sen Huang | Personal Website posenhuang.github.io. Checked 20 October 2026.
  2. Po-Sen Huang (Google): natural language processing — Lacuna lacuna.tiptreesystems.com. Checked 20 October 2026.
  3. Learning Deep Structured Semantic Models for Web Search using Clickthrough Data - Microsoft Research microsoft.com. Checked 20 October 2026.
  4. [1502.04149] Joint Optimization of Masks and Deep Recurrent Neural Networks for Monaural Source Separation arxiv.org. Checked 20 October 2026.
  5. [2112.11446] Scaling Language Models: Methods, Analysis & Insights from Training Gopher arxiv.org. Checked 20 October 2026.
  6. [2112.04359] Ethical and social risks of harm from Language Models arxiv.org. Checked 20 October 2026.
  7. [2203.07814] Competition-Level Code Generation with AlphaCode arxiv.org. Checked 20 October 2026.
  8. Scalable watermarking for identifying large language model outputs | Nature nature.com. Checked 20 October 2026.
  9. Collaboration between clinicians and vision–language models in radiology report generation | Nature Medicine nature.com. Checked 20 October 2026.
  10. AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms — Google DeepMind deepmind.google. Checked 20 October 2026.
  11. [2506.13131] AlphaEvolve: A coding agent for scientific and algorithmic discovery arxiv.org. Checked 20 October 2026.