• Invited to speak at the TCS4All Rising Star session at STOC 2026
• Co-organized Workshop on Understanding LLMs via a TCS Lens at STOC 2026
• Co-organizing a reading group, called REFORM, at Stanford
• Co-organized the Workshop on Reliable ML from Unreliable Data at NeurIPS 2025
• Our paper "What Makes Treatment Effects Identifiable?..." received the Best Paper Award at COLT 2025

Preprints

• On Language Generation in the Limit with Bounded Memory
  

  with Jon Kleinberg, Amin Saberi, and Grigoris Velegkas
  Preprint 2026

• Surprises in Proper Positive-Only Learning
  

  with Shai Ben-David, Farnam Mansouri, and Manolis Zampetakis
  Preprint 2026

• Reasoning with Sampling: Cutting at Decision Points
  

  with Quanquan C. Liu and Felix Zhou
  Preprint 2026   [Blog]

• Improved Guarantees for Heterogeneous Treatment Effect Estimation via Matrix Completion
  

  with Phuc Tran, Van H. Vu, and Manolis Zampetakis
  Preprint 2026

• What is Learnable in Valiant's Theory of the Learnable?
  

  with Steve Hanneke, Grigoris Velegkas, and Manolis Zampetakis
  Preprint 2026

Publications

• 2026
• Learning-Enabled Estimation: Tight Characterizations under Sample Selection Biases
  

  with Vikram Kher, Jane H. Lee, and Manolis Zampetakis
  EC 2026   (arXiv version coming soon!)

• Differentially Private Language Generation and Identification in the Limit
  

  with Grigoris Velegkas, Xifan Yu, and Felix Zhou
  COLT 2026

• Language Generation with Infinite Contamination
  

  with Grigoris Velegkas, Xifan Yu, and Felix Zhou
  COLT 2026

• Can SGD Select Good Fishermen? Local Convergence Under Self-Selection Biases and Beyond
  

  with Alkis Kalavasis and Felix Zhou
  COLT 2026

• DDPM Score Matching and Distribution Learning
  

  with Sinho Chewi, Alkis Kalavasis, and Omar Montasser
  COLT 2026 [Slides by Sinho] [Slides by Anay]
  Best Short Paper at ICLR 2025 Workshop on Deep Generative Model in Machine Learning

• Linear Regression with Unknown Truncation Beyond Gaussian Features
  

  with Alexandros Kouridakis, Alkis Kalavasis, and Constantine Caramanis
  ICML 2026

• Language Generation with Feedback: Queries and Mistakes
  

  with Steve Hanneke, Amin Karbasi, and Grigoris Velegkas
  ICML 2026

• Smoothed Analysis of Learning from Positive Samples
  

  with Jane H. Lee and Manolis Zampetakis
  STOC 2026

• Mean Estimation from Coarse Data: Characterizations and Efficient Algorithms
  

  with Alkis Kalavasis, Manolis Zampetakis, Felix Zhou, and Ziyu Zhu
  ICLR 2026

• Characterizations of Language Generation With Breadth
  

  with Alkis Kalavasis and Grigoris Velegkas
  ALT 2026
  Also presented at Workshop on Reliable and Responsible Foundation Models at ICML 2025

• 2025
• On Union-Closedness of Language Generation
  

  with Steve Hanneke, Amin Karbasi, and Grigoris Velegkas
  NeurIPS 2025

• What Makes Treatment Effects Identifiable? Characterizations and Estimators Beyond Unconfoundedness
  

  with Yang Cai, Alkis Kalavasis, Katerina Mamali, and Manolis Zampetakis
  COLT 2025 [COLT Talk (10 mins)] [General-audience summary] [1hr talk @ FDS, Yale]
  Best Paper Award at COLT (2025) and Invited Presentation at IJCAI (2026)

• On the Limits of Language Generation: Trade-Offs Between Hallucination and Mode Collapse
  

  with Alkis Kalavasis and Grigoris Velegkas
  STOC 2025 [Tweet] [Short Talk]
  Highlights Track at Foundations of Responsible Computing FORC (2026)

• 2024
• Tree of Attacks: Jailbreaking Black-Box LLMs Automatically
  

  with Manolis Zampetakis, Paul Kassianik, Blaine Nelson, Hyrum Anderson, Yaron Singer, and Amin Karbasi
  NeurIPS 2024
  Covered by [WIRED] [TechCrunch] · Used to safety-test Gemini and MAI-Thinking-1 Also presented at the Workshop on the Next Generation of AI Safety at ICML 2024
  Held-out data from paper is available here

• Efficient Statistics With Unknown Truncation, Polynomial Time Algorithms, Beyond Gaussians

  with Jane H. Lee and Manolis Zampetakis
  FOCS 2024  [Slides]
  Talks by Jane (20 min) and Me (1 hr)

• Smaller Confidence Intervals From IPW Estimators via Data-Dependent Coarsening

  with Alkis Kalavasis and Manolis Zampetakis
  COLT 2024  [Slides]

• Fair Classification with Partial Feedback: An Exploration-Based Data-Collection Approach

  with L. Elisa Celis and Vijay Keswani
  ICML 2024
  Also accepted for presentation at EAAMO 2024

• 2023
• Bias in Evaluation Processes: An Optimization-Based Model

  with L. Elisa Celis, Amit Kumar, and Nisheeth K. Vishnoi
  NeurIPS 2023 

• Sampling Individually-Fair Rankings that are Always Group Fair

  with Amit Deshpande, Sruthi Gorantla, and Anand Louis
  AIES 2023  [Poster]

• Subset Selection Based On Multiple Rankings in the Presence of Bias: Effectiveness of Fairness Constraints for Multiwinner Voting Score Functions

  with Niclas Boehmer, L. Elisa Celis, Lingxiao Huang, and Nisheeth K. Vishnoi
  ICML 2023 

• Maximizing Submodular Functions for Recommendation in the Presence of Biases

  with Nisheeth K. Vishnoi
  The ACM Web Conference (formerly WWW) 2023 

• 2022
• Fair Ranking with Noisy Protected Attributes

  with Nisheeth K. Vishnoi
  NeurIPS 2022  [Code] [Poster]

• Revisiting Group Fairness Metrics: The Effect of Networks

  with L. Elisa Celis and Jeffrey Sachs
  CSCW 2022 

• Selection in the Presence of Implicit Bias: The Advantage of Intersectional Constraints

  with Bary S. R. Pradelski and Nisheeth K. Vishnoi
  ACM FAccT 2022 

• Fairness for AUC via Feature Augmentation

  with Hortense Fong, Vineet Kumar, and Nisheeth K. Vishnoi
  ACM FAccT 2022 

• 2021
• Fair Classification with Adversarial Perturbations

  with L. Elisa Celis and Nisheeth K. Vishnoi
  NeurIPS 2021  [Code] [Poster]

• Mitigating Bias in Set Selection with Noisy Protected Attributes

  with L. Elisa Celis
  ACM FAccT 2021  [Code] [Poster]

• The Effect of the Rooney Rule on Implicit Bias in the Long Term

  with L. Elisa Celis, Chris Hays, and Nisheeth K. Vishnoi
  ACM FAccT 2021 

• 2019-2020
• Interventions for Ranking in the Presence of Implicit Bias

  with L. Elisa Celis and Nisheeth K. Vishnoi
  ACM FAT* 2020  [Code]

• Towards Controlling Discrimination in Online Ad Auctions

  with L. Elisa Celis and Nisheeth K. Vishnoi
  ICML 2019  [Code] [Demo] [Talk]
  Best Student Paper in the Workshop on Mechanism Design for Social Good 2019

Note

• A Note on Non-Negative L₁-Approximating Polynomials

  with Jane H. Lee and Manolis Zampetakis
  Note, 2026