Bhrij Patel | University of Maryland, College Park

Hello!

I am Bhrij (“Bridge”) Patel. I am a third-year CS PhD student at the UMD - College Park. I am co-advised by Dr. Dinesh Manocha as part of the GAMMA lab and by Dr. Amrit Singh Bedi (UCF). My work focuses on connecting theory and practice for principled policy optimization for analyzing, advancing, and aligning learning-based agents and systems. Before my PhD, I graduated from Duke University (‘22) with a double major in CS and Math and a minor in Creative Writing. During my undergrad, I was advised by Dr. Cynthia Rudin in her Interpretable ML Lab.

My current research interests:

Alignment and Personalization for AI Systems, including Multi-Agent Systems
LLM-based Evaluators, particularly for Test-time Refinement
Sample Efficient Optimization for Reinforcement Learning with Sparse Rewards

🔍 I am currently looking for research or industry internships for Spring, Summer, and Fall 2025 in the areas of Machine Learning, Reinforcement Learning, and/or LLM Systems. Feel free to reach out!

Education

Ph.D. in Computer Science, August 2022 - Present
University of Maryland, College Park

B.S. in Computer Science & Math, Minor in Creative Writing, August 2018 - May 2022
Duke University

Publications

ICML '24

Towards Global Optimality for Practical Average Reward Reinforcement Learning without Mixing Time Oracles

Bhrij Patel, Wesley A. Suttle, Alec Koppel, Vaneet Aggarwal, Brian M. Sadler, Dinesh Manocha, Amrit Singh Bedi

International Conference on Machine Learning (ICML), 2024.

PDF Code Poster

ICML '23

Beyond Exponentially Fast Mixing in Average-Reward Reinforcement Learning via Multi-Level Monte Carlo Actor-Critic

Wesley A. Suttle*, Amrit Singh Bedi*, Bhrij Patel, Brian M. Sadler, Alec Koppel, Dinesh Manocha

International Conference on Machine Learning (ICML), 2023.

PDF Code Poster

Preprints In Review

ArXiv

AIME - AI System Optimization via Multiple LLM Evalautors

Bhrij Patel, Souradip Chakraborty, Wesley A. Suttle, Mendgi Wang, Amrit Singh Bedi*, Dinesh Manocha*

PDF Code Thread

ArXiv

Multi-LLM QA with Embodied Exploration

Bhrij Patel, Vishnu Sashank Dorbala, Amrit Singh Bedi, Dinesh Manocha

PDF Code

ArXiv

Right Place, Right Time! Towards ObjectNav for Non-Stationary Goals

Vishnu Sashank Dorbala*, Bhrij Patel*, Amrit Singh Bedi, Dinesh Manocha

PDF Code

ArXiv

CCE - Sample Efficient Sparse Reward Policy Learning for Robotic Navigation via Confidence-Controlled Exploration

Bhrij Patel, Kasun Weerakoon, Wesley A. Suttle, Alec Koppel, Brian M. Sadler, Tianyi Zhou, Amrit Singh Bedi, Dinesh Manocha

PDF Code