Gaurav Sinha
 |
About MeI am a Principal Research Manager at Microsoft Research. Prior to this, I was a Senior Research Scientist at Adobe Research (2018-2022) and an Applied Scientist at Microsoft Ads (2016-2018). I received my Ph.D. in Mathematics from the California Institute of Technology in 2016, where I was advised by Prof. Eric Rains, and my Integrated M.Sc. in Mathematics and Scientific Computing from Indian Institute of Technology (IIT) Kanpur in 2011. My Ph.D. research focused on theoretical problems in mathematics and computer science. Since moving to industry, I have focused on developing principled solutions to industrial research problems, combining theory and applied research to deliver practical impact. I can be reached via email at sinhagaur88 AT gmail(dot)com. |
Research Interests
Fine Tuning and Alignment of Language Models (LMs)
Reinforcement Learning (RL) Theory and Applications (currently focussing on RL fine-tuning of LMs)
Generative Information Retrieval
Causal Machine Learning
Selected Papers (View all)
Contextual Slate GLM Bandits with Limited Adaptivity
Joint with Tanmay Goyal and Sukruta Prakash Midigeshi
ICML 2026, ArXiv
Improving Reachability on Reasoning Puzzles
Joint with Sukruta Prakash Midigeshi, Sai Soumya Nalli, Utkarsh Tiwari, Amit Deshpande, Nagarajan Natarajan, Vineeth N Balasubramanian and Amit Sharma
ICLR 2026 Workshop on Logical Reasoning of Large Language Models
Scaling the Vocabulary of Non-autoregressive Models for Fast Generative Retrieval
Joint with Ravisri Valluri, Akash Kumar Mohankumar, Kushal Dave, Amit Singh, Jian Jiao and Manik Varma
KDD 2025, ArXiv
Achieving Limited Adaptivity for Multinomial Logistic Bandits
Joint with Sukruta Prakash Midigeshi and Tanmay Goyal
RLC 2025, ArXiv
Efficient Algorithms for Logistic Contextual Slate Bandits with Bandit Feedback
Joint with Tanmay Goyal
UAI 2025, ArXiv
PlanRAG: Efficient Test-Time Planning for Retrieval Augmented Generation
Joint with Prakhar Verma, Sukruta Prakash Midigeshi, Arno Solin, Nagarajan Natarajan and Amit Sharma
Workshop on Reasoning and Planning for Large Language Models 2025, ArXiv
Generalized Linear Bandits with Limited Adaptivity
Joint with Ayush Sawarni, Nirjhar Das and Siddharth Barman
NeurIPS 2024, ArXiv
Learning good interventions in causal graphs via covering
Joint with Ayush Sawarni, Rahul Madhavan and Siddharth Barman
UAI 2023
Combinatorial categorized bandits with expert rankings
Joint with Sayak Ray Chowdhury, Nagarajan Natarajan and Amit Sharma
UAI 2023
A Causal Bandit Approach to Learning Good Atomic Interventions in Presence of Unobserved Confounders
Joint with Aurghya Maiti and Vineet Nair
UAI 2022, ArXiv
Almost Optimal Universal Lower Bound for Learning Causal DAGs with Atomic Interventions
Joint with Vibhor Porwal and Piyush Srivastava
AISTATS 2022 (Oral), ArXiv
Disentangling Mixtures of Unknown Causal Interventions
Joint with Abhinav Kumar
UAI 2021 (Oral)
Budgeted and Non-Budgeted Causal Bandits
Joint with Vineet Nair and Vishakha Patil
AISTATS 2021, ArXiv