PeopleofSuperintelligence

Researcher

Francesco Di Giovanni

Francesco Di Giovanni is a machine learning researcher whose work spans graph neural networks and generative models for chemistry and biology. After completing a mathematics PhD on Ricci flow at UCL in 2021, he coauthored the ICLR 2022 graph-curvature paper on over-squashing that received an Outstanding Paper Honorable Mention and NeurIPS 2022 work on neural sheaf diffusion. From February 2024 to June 2024, he worked as a Senior Research Associate at the University of Oxford under Michael Bronstein, and in 2024 he published a Transactions on Machine Learning Research study on how over-squashing affects the expressive power of message-passing networks. In 2026, he coauthored the MarS-FM protein simulation model and the AquaGen explicit-solvent generative model for molecular configurations. He is a Senior Research Scientist at Valence Labs, where he works on deep learning frameworks for computational methods in chemistry and biology.

Now at
Valence Labs
Badges

Timeline

  1. 2021

    A mathematics PhD at UCL

    He completes a PhD in Mathematics at UCL in 2021 under Jason Lotay, with a thesis on Ricci flow. In 2021, he publishes a study on rotationally symmetric Ricci flow on R^n+1 in Advances in Mathematics.

  2. 2022

    Curvature and graph bottlenecks

    At ICLR in 2022, he coauthors a paper with Jake Topping, Benjamin Paul Chamberlain, Xiaowen Dong and Michael M. Bronstein that introduces an edge-based combinatorial curvature to explain how negatively curved edges and bottlenecks produce over-squashing in graph neural networks. The paper receives an ICLR 2022 Outstanding Paper Honorable Mention.

  3. 2022

    Neural sheaf diffusion at NeurIPS

    At NeurIPS in 2022, he coauthors a paper on neural sheaf diffusion with Cristian Bodnar, Benjamin Chamberlain, Pietro Liò and Michael Bronstein, giving a topological perspective on heterophily and oversmoothing in graph neural networks.

  4. 2023

    Topology and over-squashing at ICML

    In an ICML 2023 paper, he and his coauthors show that graph topology plays the greatest role in over-squashing in message-passing neural networks, occurring between nodes at high commute time. At ICML in 2023, he also coauthors DRew on dynamically rewired message passing with delay.

  5. February 2024–June 2024

    Senior Research Associate at Oxford

    In February 2024, he works as a Senior Research Associate in the Department of Computer Science at the University of Oxford under Michael Bronstein; before joining Oxford, he serves as a Research Associate at the University of Cambridge in Pietro Liò's lab and as a Machine Learning Researcher at Twitter. He leaves the Oxford post in June 2024.

  6. 2024

    The expressive power of GNNs

    In 2024, he and his coauthors publish a Transactions on Machine Learning Research study on how over-squashing affects the power of graph neural networks, analyzing which function classes of node features a message-passing neural network of a given capacity can learn.

  7. April 2026

    MarS-FM for protein simulation

    In April 2026, he coauthors work on MarS-FM, a flow-based generative model for protein simulation that runs at 600 times the speed of standard molecular dynamics and generalizes to unseen protein domains.

  8. As of October 2026

    Senior Research Scientist at Valence Labs

    As of October 2026, he is a Senior Research Scientist at Valence Labs, working on deep learning frameworks for computational methods in chemistry and biology. In July 2026, he coauthors AquaGen, an all-atom explicit-solvent generative model that produces molecular configurations from the Boltzmann distribution at a fraction of the cost of molecular dynamics.

Career

  1. NowValence LabsSenior Research Scientist
  2. 2024University of OxfordSenior Research Associate
  3. University of CambridgeResearch Associate
  4. TwitterMachine Learning Researcher

Education

  1. 2021University College LondonPhD in Mathematics

From the biography.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Recursion. 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.
  2. The median h-index is 48 at universities and 14 at frontier AI labsIncludes University of Oxford. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes University of Oxford. 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. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorateIncludes University of Oxford. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorate. Across 2,116 people with both an undergraduate degree and a doctorate recorded, we trace which undergraduate schools feed which doctoral programs, how often students stay at the same school, and how paths differ by birth-country region.

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 Francesco Di Giovanni yet. Send us a free-licensed photo.

  1. I-X Seminar Series: Understanding message passing: limitations of the paradigm and new possibilities with Francesco Di Giovanni | Events | Imperial College London imperial.ac.uk. Checked 22 October 2026.
  2. Publications - Francesco Di Giovanni francescodgv.github.io. Checked 22 October 2026.
  3. Announcing the ICLR 2022 Outstanding Paper Award Recipients – ICLR Blog blog.iclr.cc. Checked 22 October 2026.
  4. Understanding over-squashing and bottlenecks on graphs via curvature arxiv.org. Checked 22 October 2026.
  5. On Over-Squashing in Message Passing Neural Networks: The Impact of Width, Depth, and Topology arxiv.org. Checked 22 October 2026.
  6. About me - Francesco Di Giovanni francescodgv.github.io. Checked 22 October 2026.
  7. Francesco Di Giovanni | Department of Computer Science cs.ox.ac.uk. Checked 22 October 2026.
  8. Publications – Computational Applied Mathematics & AI Lab camail.org. Checked 22 October 2026.
  9. Building the Virtual Protein valencelabs.substack.com. Checked 22 October 2026.
  10. AquaGen: Scaling generative models to molecular dynamics precision on thousands of atoms arxiv.org. Checked 22 October 2026.