Tommaso D'Orsi

I am a Research Scientist at Google based in New York. My work focuses on training foundation models and developing math and science data to improve their reasoning. My research also spans algorithm design, computational complexity, learning theory, and privacy.

Before joining Google, I was an Assistant Professor in the department of Computing Sciences at Bocconi. I received my PhD from ETH Zurich, where David Steurer was my advisor. I was also briefly a Research Fellow at Bocconi, hosted by Luca Trevisan.

Recent & Representative Papers [All Papers]

• Strongly Refuting Random CSP without Literals [arXiv]
  with Siu On Chan, Jeff Xu,
  FOCS 2026.
• Tight Differentially Private PCA via Matrix Coherence [arXiv]
  with Gleb Novikov,
  SODA 2026.
• Sparsest cut and eigenvalue multiplicities on low degree Abelian Cayley graphs [arXiv]
  with Chris Jones, Jake Ruotolo, Salil Vadhan and Jiyu Zhang,
  APPROX 2025, invited to the special issue of Theory of Computing.
• Private graphon estimation via sum-of-squares [arXiv]
  with Hongjie Chen, Jingqiu Ding, Yiding Hua, Chih-Hung Liu and David Steurer,
  STOC 2024.
• Private estimation algorithms for stochastic block models and mixture models [arXiv]
  with Hongjie Chen, Vincent Cohen-Addad, Alessandro Epasto, Jacob Imola, David Steurer, and Stefan Tiegel,
  NeurIPS 2023 (spotlight).
• Higher degree sum-of-squares relaxations robust against oblivious outliers [arXiv]
  with Rajai Nasser, Gleb Novikov and David Steurer,
  SODA 2023.
• Fast algorithm for overcomplete order-3 tensor decomposition [arXiv]
  with Jingqiu Ding, Chih-Hung Liu, David Steurer and Stefan Tiegel,
  COLT 2022.
• Robust Recovery for Stochastic Block Models [arXiv]
  with Jingqiu Ding, Rajai Nasser and David Steurer,
  FOCS 2021.
• Sparse PCA: Algorithms, Adversarial Perturbations and Certificates [arXiv]
  with Pravesh Kothari, Gleb Novikov and David Steurer,
  FOCS 2020.