Dr Debasis Ganguly he/him

My research passion lies at the intersection of Information Retrieval (IR) and Large Language Models (LLMs), with a particular interest in in-context learning (ICL), ranking fairness, and retrieval-augmented generation (RAG). My group explores how retrieval signals interact with generative components, aiming to build more interpretable, efficient, and equitable systems. Our recent work has delved into diverse topics such as predicting optimal example selection in ICL, understanding propagation of relevance in RAG, and using prompting to evaluate code correctness or detect misinformation.

We thrive in collaborative, interdisciplinary environments. I have co-authored with researchers in NLP, HCI, software engineering, and legal informatics, and I’m eager to extend these collaborations further. I’d particularly welcome partnerships with cognitive science, philosophy of AI, and ethics to critically examine the foundations and impacts of generative models, as well as with domain experts in medicine or law to design real-world evaluation frameworks.

CDT projects I’d be excited to supervise include: predicting performance of AI models, adaptive, fair, interpretable and trustworthy and privacy-preserving AI models, and adapting generative AI for learning. I support students exploring both theoretical depth and applied innovation, and I encourage reflective thinking about the socio-technical implications of their work.

As a supervisor, I aim to be approachable, responsive, and intellectually supportive. My goal is to help students build both independence and confidence. Several of my former students have gone on to research roles in academia and industry, contributing to leading conferences like SIGIR, ACL, and ECIR.

Equity, diversity, and inclusion are foundational to my mentoring and teaching philosophy. I have co-organized multilingual and low-resource IR shared tasks (e.g., FIRE), and contributed to workshops promoting LLM interpretability and fairness. I actively advocate for diverse representation in IR research and strive to create a welcoming, inclusive environment in all my collaborations.

Outside work, I playing chess and writing science fiction, which often inspires me to think beyond today’s technical constraints. I see my research as a form of storytelling—where the narrative is shaped not just by models and data, but also by imagination, responsibility, and curiosity.