Anay Mehrotra

Anay Mehrotra

anaymehrotra1@gmail.com

I am a Motwani Postdoctoral Fellow at Stanford hosted by Amin Saberi. I completed my Ph.D. at Yale University where I was fortunate to be advised by Amin Karbasi and Manolis Zampetakis.

I study the foundations of machine learning, especially when standard assumptions about data or learning objectives break down. My work spans learning with missing data, understanding generative models, and developing algorithms to improve the reasoning and safety of AI systems.

My work has been awarded the COLT Best Paper Award, used for red-teaming production LLMs, and featured in WIRED. I was recognized as a Rising Star in Theoretical Computer Science (by TCS4All at STOC) and a Rising Star in AI (by Oxford and Warwick’s Foundations of AI program).

Here is my Curriculum Vitae.

Recent News
Learning Theory in the Wild: Foundations of Missing Data and Language Generation

Ph.D. Thesis Yale University (2026)

Research Themes

Learning with Missing Data

What can we learn from missing or selectively observed data, and when do we need new methods?

Related activity: Workshop at NeurIPS 2025 on Reliable ML from Unreliable Data

Algorithms for Language Models

How can algorithms help test the safety of LLMs and improve their reasoning abilities?

Impact: TAP, an automated red-teaming method, was used to safety-test Gemini 2.5 and MAI-Thinking-1 and was covered by WIRED.

Related activity:

Workshop at STOC 2026 on Understanding LLMs via a TCS Lens

Reading group on Rethinking Foundations of Real-world ML (REFORM) at Stanford

Foundations of Generative Models

What does successful generation mean? When can one learn to generate, and do we face any fundamental limitations?

  • Hallucination–mode-collapse tradeoffs (STOC 2025; FORC'26 Highlights), generation with bounded memory (SODA 2027), and contamination (COLT 2026)
  • Why do diffusion models learn distributions so well?COLT 2026 (Best Short Paper at ICLR'25 workshop)

Related activity:

Tutorial at COLT 2025 on Language Generation

Workshop at Stanford on Language Generation

✦ ✦ ✦

Selected Awards

  • Rising Star in AI, Universities of Oxford and Warwick (FAIRS, 2026)
  • Rising Star in TCS, TCS4All at STOC (2026)
  • Best Paper Award, COLT (2025)
  • Sri Binay Kumar Sinha Award, IIT Kanpur (2020)
  • Finalist, ICPC World Finals (Rank 33; 2019)
  • Winner, ICPC Asia West Continent Final (2019)
✦ ✦ ✦

Service and Organization

Co-organized