I am an Applied Scientist on the frontier AI team at Amazon, where I work on LLM judges, creativity, and generalizability of LLM responses, and evaluation against other frontier models. My recent focus has been on rubrics as rewards, designing rubric-based reward signals, training rubric judges, and building training pipelines, evaluation frameworks, and dataset building methods around them.

I completed my Ph.D. in the Department of Computer Science at Purdue University in 2025, advised by Prof. Dan Goldwasser, working on contextualized political text understanding and cultural context grounding for conversations. Along the way, I interned at Microsoft Research and Amazon Alexa, and worked on the DARPA CCU program. Before Purdue, I was a Research Assistant at IIT Bombay's CFILT lab and completed my undergraduate studies at IIT Kharagpur.

Of late, I am also building a deeper foundation in the statistical theory underlying LLMs and reinforcement learning, alongside the systems engineering behind efficient, scaled-up training on parallelized accelerated compute such as GPU clusters. Within this, I am particularly drawn towards learning techniques that construct an understanding of the world through interaction and experience. I am still early in this journey and actively building up this expertise.