PeopleofSuperintelligence

Researcher

Dongdong Chen

Dongdong Chen is a computer vision researcher whose work spans neural style transfer, image restoration, vision models and multimodal AI. He earned a PhD through the University of Science and Technology of China's joint program with MSRA and coauthored StyleBank in 2017. He joined Microsoft in May 2019 and coauthored the CSWin Transformer and Florence foundation model in 2021. In 2025, he coauthored Olympus, a router for computer vision tasks, and FreeFlux, a training-free image-editing framework. As of October 2026, he leads foundational data science and horizontal post-training as a Member of Technical Staff and manager at Microsoft AI SuperIntelligence.

Now at
Microsoft

Timeline

  1. By 2019

    A joint PhD and StyleBank

    He completes a PhD at the University of Science and Technology of China (USTC) in the joint PhD program with MSRA, advised by Gang Hua, Lu Yuan and Nenghai Yu. In 2017, he coauthors the CVPR paper StyleBank with those advisors and Jing Liao; it gives each image style its own convolution filter bank atop a single auto-encoder.

  2. May 2019

    Joining Microsoft Cloud + AI

    He joins Microsoft in May 2019 after five years there as a research intern. He works in Microsoft Cloud + AI as a senior researcher.

  3. 2020

    Bringing old photos back

    In 2020, he coauthors Bringing Old Photos Back to Life with Ziyu Wan, Bo Zhang, Pan Zhang, Dong Chen, Jing Liao and Fang Wen. The CVPR 2020 oral paper trains two variational autoencoders to map old and clean photos to separate latent spaces, then learns their translation from synthetic paired data.

  4. July 2021

    The CSWin Transformer

    He coauthors the CSWin Transformer, which appears on arXiv in July 2021 and at CVPR 2022 as a vision backbone built on cross-shaped window self-attention. The model reaches 85.4% top-1 accuracy on ImageNet-1K without extra training data, 53.9 box AP on COCO and 52.2 mIoU on ADE20K.

  5. November 2021

    The Florence foundation model

    In November 2021, he is the second author of Florence, a computer vision foundation model. It adapts across classification, retrieval, object detection, visual question answering, image captioning, video retrieval and action recognition; it reports new state-of-the-art results on most of 44 benchmarks, including 83.74% top-1 accuracy in zero-shot ImageNet-1K classification.

  6. 2025

    Olympus routes computer vision tasks

    At CVPR 2025, he coauthors Olympus with Yuanze Lin, Yunsheng Li, Weijian Xu, Ronald Clark and Philip Torr; Microsoft is listed as his affiliation. Its controller multimodal large language model delegates over 20 specialized tasks across images, videos and 3D objects to dedicated modules.

  7. 2025

    FreeFlux for image editing

    At ICCV 2025, he coauthors FreeFlux with Tianyi Wei, Yifan Zhou and Xingang Pan. It analyzes layer-specific roles in RoPE-based multimodal diffusion transformers such as FLUX, then proposes a training-free image-editing framework.

  8. As of October 2026

    Member of Technical Staff

    As of October 2026, he is a Member of Technical Staff and manager at Microsoft AI SuperIntelligence. He leads the foundational data science and horizontal post-training team.

Career

  1. Since 2019MicrosoftResearch Intern, Senior Researcher

Education

  1. University of Science and Technology of ChinaPhD
  2. Microsoft Research AsiaPhD (joint program)

From the biography.

  1. Among employers with 30+ full-time cases, OpenAI leads primary technical role H-1B pay at $310,000, and Amazon, Google and Microsoft file 57% of all casesIncludes Microsoft. Across 440,669 certified US H-1B filings from 44 employers (FY2020–FY2026), OpenAI leads primary technical role base pay at $310,000 and xAI leads overall base pay at $290,000.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Microsoft. 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. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Microsoft. 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.
  4. Google has the highest combined h-index among AI organizationsIncludes Microsoft. 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.

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 Dongdong Chen yet. Send us a free-licensed photo.

  1. Dongdong Chen at Microsoft Research microsoft.com. Checked 22 October 2026.
  2. StyleBank: An Explicit Representation for Neural Image Style Transfer (CVPR 2017 Open Access Repository) openaccess.thecvf.com. Checked 22 October 2026.
  3. [2004.09484] Bringing Old Photos Back to Life arxiv.org. Checked 22 October 2026.
  4. [2107.00652] CSWin Transformer: A General Vision Transformer Backbone with Cross-Shaped Windows arxiv.org. Checked 22 October 2026.
  5. GitHub - microsoft/CSWin-Transformer: CSWin Transformer: A General Vision Transformer Backbone with Cross-Shaped, CVPR 2022 github.com. Checked 22 October 2026.
  6. [2111.11432] Florence: A New Foundation Model for Computer Vision arxiv.org. Checked 22 October 2026.
  7. Olympus: A Universal Task Router for Computer Vision Tasks (CVPR 2025 Open Access Repository) openaccess.thecvf.com. Checked 22 October 2026.
  8. Olympus: A Universal Task Router for Computer Vision Tasks (arXiv HTML) arxiv.org. Checked 22 October 2026.
  9. FreeFlux: Understanding and Exploiting Layer-Specific Roles in RoPE-Based MMDiT for Versatile Image Editing (ICCV 2025 Open Access Repository) openaccess.thecvf.com. Checked 22 October 2026.
  10. Dongdong Chen dongdongchen.bid. Checked 22 October 2026.