My research interests are at the intersection of Algorithms, Distributed Systems, and Artificial Intelligence. I focus on building efficient, resilient, and intelligent systems.

Master's Thesis: Frontiers of LLM Reasoning

Under the supervision of Professor Rezaul Chowdhury at the Theoretical and Experimental Algorithmics Lab, my thesis investigates the logical reasoning capabilities of Large Language Models (LLMs).

My research aims to identify the "frontier" of algorithmic problems that remain challenging for LLMs due to their probabilistic nature, comparing their performance directly against human problem-solving strategies. The project involves a multi-faceted approach, including inference optimization, model fine-tuning, developing graph-based retrieval systems, and designing advanced prompting strategies. A key contribution of this work is the creation of a custom autograder to systematically benchmark and evaluate the reasoning flows of various models.

Carbon-Aware Cloud Computing

As a researcher at Pace Lab, I am developing a Linux tool that computes the carbon consumption of individual jobs in cloud data centers. This tool provides crucial data for creating intelligent, carbon-aware scheduling policies, helping cloud providers enhance hardware longevity and reduce their environmental impact.

Fault-Tolerant Distributed Databases

I am engineering a modern, fault-tolerant fork of MySQL by implementing the Raft and Practical Byzantine Fault Tolerance (PBFT) consensus algorithms. This work enhances the database's resilience and performance, making it suitable for robust, large-scale distributed applications.