PeopleofSuperintelligence

Researcher

Xiang Fu

Also known as 付襄

Xiang Fu is a researcher whose work ranges from reinforcement learning to AI-based materials design. He coauthored a 2021 method for task-informed abstractions and diffusion models for crystals and metal-organic frameworks; he also helped develop MatterGen, which generates inorganic materials from target properties. In 2025, he coauthored UMA for atomistic modeling and All-atom Diffusion Transformers for molecules and materials. As of October 2026, he is a member of technical staff at Periodic Labs.

Timeline

  1. 2021

    Learning task informed abstractions

    In 2021, he coauthors Learning Task Informed Abstractions with Ge Yang, Pulkit Agrawal, and Tommi Jaakkola at ICML. The method separates reward-correlated visual features from distractors for model-based reinforcement learning.

  2. 2021

    Crystal diffusion VAE

    In 2021, he coauthors Crystal Diffusion Variational Autoencoder for Periodic Material Generation. The model learns from the distribution of stable materials and uses diffusion to generate crystal structures, moving atomic coordinates toward lower-energy states.

  3. 2023

    MOFDiff for framework design

    In 2023, he coauthors MOFDiff, a coarse-grained diffusion model that generates metal-organic framework structures by denoising the coordinates and identities of their building blocks.

  4. January 2025

    MatterGen for materials design

    In January 2025, he coauthors MatterGen with Microsoft Research AI for Science colleagues. The model directly generates novel materials from application requirements and can target chemistry and mechanical, electronic, or magnetic properties.

  5. March 2025

    All-atom diffusion transformers

    In March 2025, he coauthors All-atom Diffusion Transformers, a latent diffusion framework that jointly generates periodic materials and non-periodic molecules.

  6. June 2025

    Universal models for atoms

    In June 2025, he coauthors UMA, Meta FAIR's family of Universal Models for Atoms. The models train on half a billion 3D atomic structures across molecules, materials, and catalysts.

  7. By 2026

    A PhD at MIT CSAIL

    By 2026, he holds a PhD from MIT CSAIL, advised by Tommi Jaakkola with additional mentorship from Pulkit Agrawal. His career includes work at Meta FAIR.

  8. As of October 2026

    Member of technical staff

    As of October 2026, he is a member of technical staff at Periodic Labs.

Career

  1. NowPeriodic LabsMember of Technical Staff

Education

  1. MIT CSAILPhD

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 Periodic Labs. 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. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Periodic Labs. 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 Periodic Labs. 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. 22 of 133 people shaping AI with a former OpenAI role now work at AnthropicIncludes Periodic Labs. 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.

Sources and credits

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

  1. Learning Task Informed Abstractions proceedings.mlr.press. Checked 21 October 2026.
  2. Crystal Diffusion Variational Autoencoder for Periodic Material Generation arxiv.org. Checked 21 October 2026.
  3. MOFDiff: Coarse-grained Diffusion for Metal-Organic Framework Design arxiv.org. Checked 21 October 2026.
  4. MatterGen: A new paradigm of materials design with generative AI microsoft.com. Checked 21 October 2026.
  5. All-atom Diffusion Transformers: Unified generative modelling of molecules and materials arxiv.org. Checked 21 October 2026.
  6. UMA: A Family of Universal Models for Atoms arxiv.org. Checked 21 October 2026.
  7. Xiang Fu xiangfu.co. Checked 21 October 2026.