PeopleofSuperintelligence

Researcher

Marco Ciccone

Marco Ciccone is a machine-learning researcher whose work spans visual representation learning, federated learning, and modular AI systems. He wrote an M.Sc. thesis at Politecnico di Milano in 2016 and pursued doctoral work there by 2021, then co-chaired the NeurIPS Competition Track from 2021 to 2023. He pursued postdoctoral research at Politecnico di Torino and University College London, coauthored FedGWC for clustered federated learning, and coauthored three ICML papers in 2026 on tokenization, low-rank deployment, and adaptive LoRA merging. As of October 2026, he is a Vector Distinguished Postdoctoral Fellow working with Colin Raffel on modular intelligent systems.

Timeline

  1. 2016

    M.Sc. thesis at Politecnico di Milano

    In 2016, he completes an M.Sc. thesis in Computer Science and Engineering at Politecnico di Milano on semantic segmentation of urban scenes with recurrent neural networks.

  2. By 2021

    Doctoral work on visual learning

    By 2021, he is a Ph.D. candidate in Computer Science at Politecnico di Milano, where he studies iterative and conditional computation for visual representation learning.

  3. 2021–2023

    Co-chairing NeurIPS competitions

    He co-chairs the NeurIPS Competition Track in 2021, 2022, and 2023. In 2023, he coauthors an announcement for 20 accepted competitions.

  4. 2022

    An ELLIS postdoctoral role

    In 2022, he works as a postdoctoral researcher at Politecnico di Torino on knowledge transfer across tasks for incremental learning and privacy-preserving applications. His ELLIS postdoctoral role also includes University College London.

  5. 2025

    FedGWC for clustered learning

    In 2025, he coauthors FedGWC, an interaction-aware method for clustered federated learning. It groups clients by their data distributions through a Gaussian reward mechanism to build more robust, personalized models.

  6. 2026

    Three ICML papers

    In 2026, he coauthors three ICML papers: TokSuite measures how tokenizer choice affects language models; FlexRank uses nested low-rank decomposition for adaptive deployment; and The Appeal and Reality of Recycling LoRAs with Adaptive Merging examines how reused adapters perform.

  7. As of October 2026

    A Vector postdoctoral role

    As of October 2026, he is a Vector Distinguished Postdoctoral Fellow at Vector Institute, where he works with Colin Raffel on modular intelligent systems that reuse, compose, and improve learned skills.

Career

  1. NowVector InstituteVector Distinguished Postdoctoral Fellow
  2. From 2022Politecnico di TorinoPostdoctoral researcher
  3. From 2022University College LondonPostdoctoral researcher

Education

  1. 2016Polytechnic University of MilanM.Sc.

From the biography.

  1. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Vector Institute. 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.
  2. Nonprofit roles are 2.3 times as common in AI safetyIncludes Vector Institute. We estimate where people focused on safety, alignment and evaluation work. Nonprofit roles appear for 25.1% of safety people with known affiliations, compared with 11.0% in other areas.
  3. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes Vector Institute. 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.
  4. The United States employs an estimated 48% of people shaping AIIncludes Vector Institute. 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.

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 Marco Ciccone yet. Send us a free-licensed photo.

  1. Marco Ciccone marcociccone.github.io. Checked 21 October 2026.
  2. Marco Ciccone marcociccone.github.io. Checked 21 October 2026.
  3. NeurIPS 2021 Call for Competitions blog.neurips.cc. Checked 21 October 2026.
  4. NeurIPS 2022 Call for Competitions blog.neurips.cc. Checked 21 October 2026.
  5. Announcing the NeurIPS 2023 Competitions blog.neurips.cc. Checked 21 October 2026.
  6. Interaction-Aware Gaussian Weighting for Clustered Federated Learning arxiv.org. Checked 21 October 2026.
  7. TokSuite: Measuring the Impact of Tokenizer Choice on Language Model Behavior arxiv.org. Checked 21 October 2026.
  8. FlexRank: Nested Low-Rank Knowledge Decomposition for Adaptive Model Deployment arxiv.org. Checked 21 October 2026.
  9. The Appeal and Reality of Recycling LoRAs with Adaptive Merging arxiv.org. Checked 21 October 2026.
  10. Marco Ciccone - Vector Institute for Artificial Intelligence vectorinstitute.ai. Checked 21 October 2026.