PeopleofSuperintelligence

Researcher

Ning Zhang

Ning Zhang is a computer vision researcher whose work on deep features and fine-grained recognition at UC Berkeley now extends to multimodal models at Meta. After graduating from Tsinghua University in 2010, they earned a PhD at Berkeley in 2015 under Trevor Darrell and co-authored DeCAF, PANDA and Part-based R-CNNs along the way. In 2024, DeCAF received the ICML Test of Time Award, and Zhang was among the authors of The Llama 3 Herd of Models. Earlier roles included heading computer vision at Dawnlight and leading the computer vision research group at Snapchat. Zhang is a senior research scientist manager at Meta AI, working on computer vision, language and multimodal models for commerce and monetization.

Now at
Meta
Badges

Timeline

  1. 2010

    Tsinghua computer science degree

    In 2010, Zhang graduates from Tsinghua University with a B.S. in computer science, working with Professor Jie Tang.

  2. 2013

    DeCAF deep convolutional features

    The DeCAF paper, posted on arXiv in 2013 with Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Eric Tzeng and Trevor Darrell, tests whether features from a deep convolutional network trained for object recognition can be reused for new tasks. It releases an open-source implementation of the features together with the network parameters.

  3. 2013–2014

    PANDA and Part-based R-CNNs

    In 2013, Zhang co-authors PANDA with Manohar Paluri, Marc'Aurelio Ranzato, Trevor Darrell and Lubomir Bourdev, a method that trains pose-normalized convolutional networks to infer human attributes such as gender and clothes style. Part-based R-CNNs for Fine-grained Category Detection follows in 2014, with Zhang as first author alongside Jeff Donahue, Ross Girshick and Trevor Darrell; the model learns whole-object and part detectors and beats earlier fine-grained methods on the Caltech-UCSD bird dataset without a bounding box at test time.

  4. 2015

    PhD with Trevor Darrell

    In 2015, Zhang earns a PhD in computer science at UC Berkeley, advised by Trevor Darrell, with the dissertation Visual Representations for Fine-grained Categorization. The thesis covers pose-normalized representations, Part-based R-CNNs and a fully convolutional network trained end to end for part localization and fine-grained classification. Beyond Frontal Faces, a 2015 paper with Manohar Paluri, Yaniv Taigman, Rob Fergus and Lubomir Bourdev, introduces the PIPA dataset of over 60000 instances of 2000 individuals and the PIPER method for person recognition.

  5. 2022

    CommerceMM multimodal model

    In 2022, Zhang co-authors CommerceMM with Licheng Yu, Jun Chen, Animesh Sinha, Mengjiao Wang, Hugo Chen and Tamara L. Berg. The multimodal model understands commerce content in images, text and image-plus-text, and after fine-tuning it reaches state-of-the-art results on 7 commerce-related downstream tasks.

  6. 2024–2025

    Llama 3 and Apollo

    In 2024, Zhang is among the authors of The Llama 3 Herd of Models, a set of foundation models whose largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. The paper Apollo, another Zhang co-authorship, studies what drives video understanding in large multimodal models, introduces the Apollo model family, and appears at CVPR in 2025.

  7. 2024

    ICML Test of Time Award

    In 2024, the ICML Test of Time Award goes to DeCAF, a 2014 paper whose co-authors include Zhang, Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Eric Tzeng and Trevor Darrell. The award celebrates the lasting impact of a paper over the past 10 years, and its ceremony takes place in Vienna, Austria.

  8. As of October 2026

    Research manager at Meta AI

    Zhang is a senior research scientist manager at Meta AI, working on computer vision, natural language processing and multimodal models for commerce and monetization applications. Earlier roles include heading computer vision at Dawnlight, where the team built a first real-time activity recognition prototype for edge devices, and leading the computer vision research group at Snapchat. The service record includes Area Chair roles at CVPR in 2021, 2022 and 2023 and Diversity and Inclusion Chair at ICCV in 2021.

Career

  1. NowMetaSenior research scientist manager
  2. DawnlightHead of Computer Vision
  3. SnapchatComputer vision research group lead

Education

  1. 2015University of California, BerkeleyPhD
  2. 2010Tsinghua UniversityB.S.

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Ning Zhang. 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 company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Meta. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  4. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Meta. 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.

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 Ning Zhang yet. Send us a free-licensed photo.

  1. Ning Zhang's homepage n-zhang.github.io. Checked 22 October 2026.
  2. DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition (arXiv) arxiv.org. Checked 22 October 2026.
  3. PANDA: Pose Aligned Networks for Deep Attribute Modeling (arXiv) arxiv.org. Checked 22 October 2026.
  4. Part-based R-CNNs for Fine-grained Category Detection (arXiv) arxiv.org. Checked 22 October 2026.
  5. Visual Representations for Fine-grained Categorization | EECS at UC Berkeley www2.eecs.berkeley.edu. Checked 22 October 2026.
  6. Beyond Frontal Faces: Improving Person Recognition Using Multiple Cues (arXiv) arxiv.org. Checked 22 October 2026.
  7. CommerceMM: Large-Scale Commerce MultiModal Representation Learning with Omni Retrieval (arXiv) arxiv.org. Checked 22 October 2026.
  8. The Llama 3 Herd of Models (arXiv) arxiv.org. Checked 22 October 2026.
  9. Apollo: An Exploration of Video Understanding in Large Multimodal Models (arXiv) arxiv.org. Checked 22 October 2026.
  10. Ning Zhang - Google Scholar scholar.google.com. Checked 22 October 2026.
  11. Trevor Darrell receives ICML Test-of-Time Award - EECS at Berkeley eecs.berkeley.edu. Checked 22 October 2026.
  12. ICML Test Of Time DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition icml.cc. Checked 22 October 2026.