Lucas Rosenblatt

Link to CV

I am a fourth year PhD candidate at NYU where I am advised by Julia Stoyanovich and work closely with Christoper Musco. At NYU, I am affiliated with the NYU Center for Responsible AI and the Theoretical Computer Science @ NYU Group. I work closely with Bill Howe (of UW) and with the Volitional AI Lab (also at UW). I also work with Rachel Cummings (of Columbia) and her group. I am grateful to be supported by a NSF Graduate Research Fellowship.

Broadly, my work aims to answer open questions on data privacy, algorithmic fairness, climate, AI with social impact, and LLMs, all with an eye towards doing social good.

I was formerly a member of the Microsoft AI rotational program, working out of the New England Research and Development lab (and remotely during COVID!). In 2019 I graduated from Brown University, where I wrote a thesis about AI and self-data collection.

I happen to own a school bus that I’ve spent a lot of time converting into a mobile home and finding a permanent home for it in rural Vermont. If you like, I’ll give you some great reasons to buy a bus, and arguably some better reasons not to. I also make movies and write as much as I can.

Service

Reviewer: Neurips 2025, 2024, KDD 2025, ICML 2025, FAccT 2025, ICLR 2025, AISTATS 2025, TPDP 2025, SOSA 2024, CHI 2023

news

Jun 19, 2025	Our paper (Do You Really Need Public Data? Surrogate Public Data for Differential Privacy on Tabular Data), exploring what we can do under DP when generating public tabular data with LLMs, was selected for a Spotlight Presentation (top 10% of submissions) at the FMSD Workshop @ICML2025!
May 19, 2025	(Belatedly, and happily) sharing that we have two forthcoming papers at ICML 2025 (say hi in Vancouver!), the first, on a new (and I think well motivated) threat model for LLMs, Fragments to Facts: Partial-Information Fragment Inference from LLMs is with Bin Han, Robert Wolfe, and Bill Howe. and the second, which we had already released as a pre-print, Differential Privacy Under Class Imbalance: Methods and Empirical Insights with Rachel Cummings and Marco Avella Medina.
May 18, 2025	Very excited to (belatedly) announce that our work, Sharper Bounds for Chebyshev Moment Matching with Applications to Differential Privacy and Beyond, was accepted to COLT 2025. Say hi in Lyon : )
Apr 22, 2025	Some new work (pre-print here!) I’m really happy with that explores what we can under DP when generating public tabular data with LLMs, with Shlomi Hod and Julia Stoyanovich. This of course assumes that LLMs can truly be considered public data, which is up for debate…
Apr 21, 2025	I will join Google Research this summer as a Research Intern working on data privacy problems.
Nov 8, 2024	New pre-print with Rachel Cummings and Marco Avella Medina: Differential Privacy Under Class Imbalance: Methods and Empirical Insights. Very happy with how the work turned out, and there’s still tons to be done in this space! We presented a preliminary version of the work at TPDP 2024.
Oct 2, 2024	Teal Witter and I just posted our preprint (under submission): FairlyUncertain: A Comprehensive Benchmark of Uncertainty in Algorithmic Fairness. Really excited about some clever evaluation schemes for assessing fairness AND uncertainty under one umbrella benchmark!
Aug 11, 2024	New work that I’m very excited about, Sharper Bounds for Chebyshev Moment Matching with Applications to Differential Privacy and Beyond with Chris Musco, Cam Musco, and Apoorv Vikram Singh. Of the applications of some cool theory in this work: Chebyshev moment matching is a promising mechanism for learning private synthetic distributions! We presented an abstract of the work at TPDP 2024, and the full version is under submission.
Mar 30, 2024	Our paper Laboratory-Scale AI: Open-Weight Models are Competitive Even in Low-Resource Settings was accepted to FAccT 2024. This work was led by the indomitable Robert Wolfe; much credit to him, and happy to have been part of the team!
Feb 5, 2024	Excited to be organizing the Spring iteration of NYC Privacy Day, to be held at NYU on April 19th, 2024. Privacy Day is a seasonal event that unites privacy folks from NYC institutions; hosting rotates between Columbia, NYU, Google and Cornell Tech. RSVP through the website if you can make it!