PeopleofSuperintelligence

Researcher

Donggyun Kim

Donggyun Kim is a KAIST computer scientist whose research connects generative modeling, meta-learning and few-shot vision. At KAIST, he earned a bachelor's degree in mathematical science with a computer science double major in 2020, a computer science master's in 2022 and a doctorate in 2026. He coauthored ID-GAN in 2020 and Multi-Task Neural Processes in 2022; his Visual Token Matching paper received ICLR's 2023 Outstanding Paper Award. Chameleon applied his work to dense visual prediction in 2024, and his 2025 papers explored gradient-free adaptation and few-shot spatial control for diffusion models. After visiting NYU's Agentic AI Learning Lab, he joined KAIST's Information & Electronics Research Institute in March 2026 and works as an InnoCore postdoctoral researcher.

Now at
KAIST
Works
85
Citations
543
h-index
12

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2015–2026

    Degrees and a thesis at KAIST

    At KAIST, he earns a B.S. in Mathematical Science with a computer science double major in 2020 and an M.S. in computer science in 2022. He completes a computer science Ph.D. in February 2026 under Seunghoon Hong. His 2026 dissertation, Universal few-shot learning with efficient adaptation for generalist models, studies expanding few-shot learning and reducing the cost of adapting foundation models.

  2. 2020

    High-fidelity image synthesis

    He coauthors ID-GAN in 2020, a framework for disentangled representation and high-fidelity image synthesis.

  3. 2022

    Multi-task neural processes

    In 2022, he coauthors Multi-Task Neural Processes, which extends neural processes to infer tasks from multiple stochastic processes.

  4. 2023

    Visual Token Matching wins ICLR award

    He coauthors Visual Token Matching in 2023 as a universal few-shot learner for dense prediction. ICLR gives the paper its Outstanding Paper Award.

  5. 2024

    A vision generalist called Chameleon

    His Chameleon paper presents a data-efficient vision generalist for dense prediction in 2024, adapting to unseen label structures from a few examples.

  6. 2025

    Efficient adaptation and diffusion control

    In 2025, he proposes HyperFlow, which emulates gradient descent without computing gradients, and Universal Few-Shot Control, a diffusion-model adapter for novel spatial conditions.

  7. November 2025–February 2026

    Visiting scholar at NYU

    He visits NYU's Agentic AI Learning Lab from November 2025 to February 2026 and collaborates with Mengye Ren.

  8. Since March 2026

    Postdoctoral researcher at KAIST

    He is a postdoctoral researcher at KAIST's Information & Electronics Research Institute and an InnoCore member of the AI Meta-Scientist Research Center from March 2026.

Career

  1. Since 2026KAISTPostdoctoral Researcher
  2. 2025–2026New York UniversityVisiting Scholar
  3. 2016–2024Konkuk University
  4. 2010–2016Chung-Ang University

Education

  1. 2026KAISTPh.D.
  2. 2022KAISTM.S.
  3. 2020KAISTB.S.

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

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Donggyun Kim. 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.
  2. The United States employs an estimated 48% of people shaping AIIncludes KAIST. We estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.
  3. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes KAIST. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  4. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes KAIST. 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.

Colleagues

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

Sources and credits

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

  1. Donggyun Kim gitgyun.github.io. Checked 21 October 2026.
  2. Universal few-shot learning with efficient adaptation for generalist models library.kaist.ac.kr. Checked 21 October 2026.
  3. High-Fidelity Synthesis with Disentangled Representation ecva.net. Checked 21 October 2026.
  4. Multi-Task Neural Processes pure.kaist.ac.kr. Checked 21 October 2026.
  5. Universal Few-shot Learning of Dense Prediction Tasks with Visual Token Matching pure.kaist.ac.kr. Checked 21 October 2026.
  6. ICLR 2023 Awards iclr.cc. Checked 21 October 2026.
  7. Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild pure.kaist.ac.kr. Checked 21 October 2026.
  8. [2504.15323] HyperFlow: Gradient-Free Emulation of Few-Shot Fine-Tuning arxiv.org. Checked 21 October 2026.
  9. [2509.07530] Universal Few-Shot Spatial Control for Diffusion Models arxiv.org. Checked 21 October 2026.