PeopleofSuperintelligence

Researcher

Nan Du

Nan Du is a machine learning researcher who works on giant language models and efficient AI at OpenAI. A research assistant at Georgia Tech from 2011 to 2016, they co-wrote a paper on influence estimation in diffusion networks that won a NIPS Outstanding Paper Award in 2013, and joined Google as a Senior Research Scientist in 2016. At Google they were first author of the 1.2-trillion-parameter GLaM mixture-of-experts model in 2021 and one of the PaLM authors in 2022. They then worked in Apple's Foundation Models group, co-authoring MM1 and the Apple Intelligence language models in 2024, and went to OpenAI in 2025. A paper from their Apple years, on Parallel Track Transformers, appeared in February 2026.

Now at
OpenAI
Works
153
Badges

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2011–2016

    Research at Georgia Tech

    From 2011 to 2016, they are a research assistant in the School of Computational Science and Engineering at Georgia Tech's College of Computing. Their work covers machine learning for user-engagement modeling, online information diffusion and advertisement assignment over large social networks, and a Facebook Graduate Fellowship supports them in 2014–2015.

  2. 2013

    A NIPS Outstanding Paper Award

    In 2013, the paper “Scalable Influence Estimation in Continuous-Time Diffusion Networks,” written with Le Song, Manuel Gomez-Rodriguez and Hongyuan Zha, wins a NIPS Outstanding Paper Award.

  3. 2016

    Senior research scientist at Google

    In 2016, they publish “Recurrent Marked Temporal Point Processes: Embedding Event History to Vector” in the KDD proceedings with Hanjun Dai, Le Song and others. They join Google as a Senior Research Scientist in July 2016.

  4. 2021–2022

    GLaM and PaLM at Google

    In December 2021, as a research scientist on Google Research's Brain Team, they are first author of GLaM, a sparsely activated mixture-of-experts language model whose largest version has 1.2 trillion parameters and uses one third of the energy used to train GPT-3. In 2022, they are among the authors of PaLM, a 540-billion-parameter Transformer language model trained with the Pathways system.

  5. 2024

    Apple's MM1 and Apple Intelligence models

    In Apple's Foundation Models group, led by Ruoming Pang, they are among the authors of MM1 in March 2024, a family of multimodal models of up to 30B parameters with mixture-of-experts variants. In July 2024, they are among the authors of the Apple Intelligence Foundation Language Models report, which covers a model of about 3 billion parameters for devices and a larger server model.

  6. 2025

    From Apple to OpenAI

    In 2025, they leave Apple, where they are a principal researcher and a specialist in mixture-of-experts architectures, for OpenAI. John Peebles also goes to OpenAI, as the Foundation Models team loses roughly 10 members, including its chief.

  7. February 2026

    Parallel Track Transformers

    A paper from their time at Apple, Parallel Track Transformers, is published in February 2026 with Chong Wang, Tom Gunter, Ruoming Pang and others. The architecture restructures computation to minimize cross-device dependencies and reaches up to a 16x reduction in synchronization operations relative to standard tensor parallelism.

  8. As of October 2026

    Giant models at OpenAI

    They work at OpenAI on giant models and efficient AI. Their most-cited paper is ReAct, from 2022, written with Shunyu Yao and others on interleaving reasoning and acting in language models.

Career

  1. Since 2025OpenAI
  2. From 2024ApplePrincipal Researcher
  3. 2016–2022GoogleSenior Research Scientist, Research Scientist
  4. 2011–2016Georgia Institute of TechnologyResearch assistant

From the biography.

  1. 22 of 133 people shaping AI with a former OpenAI role now work at AnthropicIncludes Nan Du. Our estimates of where OpenAI recruits came from and where 133 people with former OpenAI roles work now, including the companies and labs they founded.
  2. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Nan Du. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  3. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Nan Du. 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.
  4. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Nan Du. 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.

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 Nan Du yet. Send us a free-licensed photo.

  1. NAN DU (personal homepage) dunan.github.io. Checked 20 October 2026.
  2. NIPS 2013 Awards neurips.cc. Checked 20 October 2026.
  3. More Efficient In-Context Learning with GLaM research.google. Checked 20 October 2026.
  4. GLaM: Efficient Scaling of Language Models with Mixture-of-Experts arxiv.org. Checked 20 October 2026.
  5. PaLM: Scaling Language Modeling with Pathways arxiv.org. Checked 20 October 2026.
  6. MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training arxiv.org. Checked 20 October 2026.
  7. Apple Intelligence Foundation Language Models arxiv.org. Checked 20 October 2026.
  8. Trapped by silence: How Apple lost its top AI talent to Meta kr-asia.com. Checked 20 October 2026.
  9. Apple's AI exodus continues with four new deflections to Meta, OpenAI and Anthropic 9to5mac.com. Checked 20 October 2026.
  10. Parallel Track Transformers: Enabling Fast GPU Inference with Reduced Synchronization machinelearning.apple.com. Checked 20 October 2026.
  11. NAN DU - Google Scholar scholar.google.com. Checked 20 October 2026.