PeopleofSuperintelligence

Researcher

Haggai Maron

Haggai Maron is a machine-learning researcher who studies deep learning on sets, graphs, point clouds and the weights of other networks, using the symmetries these objects have. After finishing a PhD under Yaron Lipman at the Weizmann Institute of Science in 2019, he joined NVIDIA Research in October 2019. In 2020, his paper On Learning Sets of Symmetric Elements, written with Or Litany, Gal Chechik and Ethan Fetaya, won an ICML Outstanding Paper Award. He joined the Technion's Faculty of Electrical and Computer Engineering as an assistant professor in 2023, received an Alon Fellowship in 2024, and received an ERC Starting Grant in 2026. He is an assistant professor at the Technion and a senior research scientist at NVIDIA Research in Tel Aviv.

Now at
NVIDIA and Technion
Badges

Timeline

  1. By 2019

    Weizmann PhD and graph networks

    He earns a BSc in mathematics and computer science at the Hebrew University of Jerusalem, then an MSc and a PhD in the Department of Computer Science and Applied Mathematics at the Weizmann Institute of Science, finishing the PhD in 2019 under Yaron Lipman. With Lipman and others, he coauthors Invariant and Equivariant Graph Networks in 2018, which characterizes all permutation invariant and equivariant linear layers for graph data, and Provably Powerful Graph Networks in 2019, which gives a simple model with provable 3-WL expressive power.

  2. October 2019

    Joining NVIDIA Research

    He joins NVIDIA Research as a research scientist in October 2019. By 2022, he is a senior research scientist and a member of NVIDIA's TLV lab.

  3. 2020

    An ICML Outstanding Paper

    In 2020, On Learning Sets of Symmetric Elements, written with Or Litany, Gal Chechik and Ethan Fetaya, receives an Outstanding Paper Award at ICML. Its Deep Sets for Symmetric elements (DSS) layers respect both the reordering of set elements and the symmetries of the elements themselves, and networks built from them are universal approximators of invariant and equivariant functions.

  4. 2021–2022

    Subgraph aggregation networks

    In 2021, he coauthors Equivariant Subgraph Aggregation Networks (ESAN) with Beatrice Bevilacqua, Fabrizio Frasca, Derek Lim, Michael Bronstein and others. ESAN represents each graph as a set of subgraphs and processes it with an equivariant architecture, which increases the expressive power of message-passing networks. In December 2022, he introduces the framework in a DEGAS webinar of the IEEE Signal Processing Society.

  5. 2023

    Technion and weight-space learning

    In 2023, he joins the Faculty of Electrical and Computer Engineering at the Technion as an assistant professor. The 2023 paper Equivariant Architectures for Learning in Deep Weight Spaces, written with Aviv Navon, Aviv Shamsian, Idan Achituve, Ethan Fetaya and Gal Chechik, presents a network that takes the weights and biases of a pre-trained MLP as input and processes them with layers equivariant to the permutation of neurons.

  6. 2024

    An Alon Fellowship

    In 2024, he receives the Alon Fellowship, a scholarship for the integration of outstanding faculty.

  7. 2026

    An ERC Starting Grant

    In 2026, he receives an ERC Starting Grant for the project Geometric Deep Learning for Neural Artifacts (GDLNeuralArt) and the ICSA Early Career Award.

  8. As of October 2026

    Technion professor and NVIDIA researcher

    He is an assistant professor in the Technion's Faculty of Electrical and Computer Engineering and a senior research scientist at NVIDIA Research in Tel Aviv. His research applies deep learning to sets, graphs, point clouds, surfaces and weight spaces, using the symmetry structure of these objects.

Career

  1. Since 2023TechnionAssistant Professor
  2. Since 2019NVIDIAResearch Scientist

Education

  1. 2019Weizmann Institute of ScienceMSc
  2. Hebrew University of JerusalemBSc

From the biography.

  1. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. 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.
  2. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  3. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. 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. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes NVIDIA. Among 819 people whose first recorded role started in 2020–2026 and who have a known current organization, 60% work in commercial sectors (25% at startups and other companies, 23% at frontier labs and 13% at big tech, with 9 in two), and 31% work at universities, compared with 66% for earlier entrants.

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 Haggai Maron yet. Send us a free-licensed photo.

  1. Haggai Maron | Research research.nvidia.com. Checked 21 October 2026.
  2. Haggai Maron haggaim.github.io. Checked 21 October 2026.
  3. Invariant and Equivariant Graph Networks arxiv.org. Checked 21 October 2026.
  4. Provably Powerful Graph Networks arxiv.org. Checked 21 October 2026.
  5. SPS DSI (DEGAS) Webinar: 14 December 2022, presented by Dr. Haggai Maron signalprocessingsociety.org. Checked 21 October 2026.
  6. ICML 2020 Awards icml.cc. Checked 21 October 2026.
  7. Interview with Haggai Maron – #ICML2020 award winner aihub.org. Checked 21 October 2026.
  8. Equivariant Subgraph Aggregation Networks arxiv.org. Checked 21 October 2026.
  9. Equivariant Architectures for Learning in Deep Weight Spaces arxiv.org. Checked 21 October 2026.