Dhruv Sarkar

ধ্রুব সরকার · ध्रुव सरकार

I am an incoming graduate student at the Massachusetts Institute of Technology, where I will be advised by Prof. Ramesh Raskar. I am broadly interested in building principled learning systems that can make reliable decisions, adapt through interaction, and align with human objectives.

I spent five years as an undergraduate at the Indian Institute of Technology Kharagpur, where I worked on the theoretical foundations of Machine Learning and Reinforcement Learning. I was fortunate to be advised by Prof. Abhishek Sinha, Prof. Sayak Ray Chowdhury, and Prof. Vaneet Aggarwal.

My current interests span sequential decision-making and interactive learning, test-time alignment of language models, generative models, stochastic optimization, and computational optimal transport. During May–July 2026, I was also a visiting student at the Mohamed bin Zayed University of Artificial Intelligence, hosted by Prof. Junpei Komiyama.

Sequential decision-making Interactive learning LLM alignment Generative models Stochastic optimization Optimal transport
CV Google Scholar Email

news

Jul 16, 2026 I will join the MIT Media Lab as a graduate student in September 2026, advised by Prof. Ramesh Raskar.
Jun 12, 2026 New preprint with Prof. Vaneet Aggarwal: Nonlinear Two-Time-Scale Stochastic Approximation: A Sharp Phase Transition and How to Beat It.
May 20, 2026 New preprint with Prof. Abhishek Sinha: Improved Guarantees for Constrained Online Convex Optimization via Self-Contraction.
May 01, 2026 I joined MBZUAI as a visiting student, hosted by Prof. Junpei Komiyama in the Department of Machine Learning.
Apr 15, 2026 Two papers accepted at ICML 2026: Optimal Anytime Algorithms for Online Convex Optimization with Adversarial Constraints and Improved Algorithms for Nash Welfare in Linear Bandits.
Feb 03, 2026 Three papers accepted at AISTATS 2026, on projection-free constrained online learning, relation-aware cross-domain alignment, and social welfare in bandits.

selected publications

  1. ICML
    Optimal Anytime Algorithms for Online Convex Optimization with Adversarial Constraints
    Dhruv Sarkar, and Abhishek Sinha
    In The 43rd International Conference on Machine Learning. Preliminary version appeared at the AAAI 2026 MATH4AI Workshop. , 2026
    arXiv Bib PDF
  2. ICML
    Improved Algorithms for Nash Welfare in Linear Bandits
    Dhruv Sarkar, Nishant Pandey, and Sayak Ray Chowdhury
    In The 43rd International Conference on Machine Learning, 2026
    arXiv Bib PDF
  3. AISTATS
    Projection-free Algorithms for Online Convex Optimization with Adversarial Constraints
    Dhruv Sarkar, Aprameyo Chakrabartty, Subhamon Supantha, Palash Dey, and Abhishek Sinha
    In The 29th International Conference on Artificial Intelligence and Statistics, 2026
    arXiv Bib HTML PDF
  4. AAAI
    DP-NCB: Privacy Preserving Fair Bandits
    Dhruv Sarkar, Nishant Pandey, and Sayak Ray Chowdhury
    In The 40th AAAI Conference on Artificial Intelligence, 2026
    DOI arXiv Bib PDF