Debrup Das

Description of image
I am a 1st year Computer Science PhD student at the Manning College of Information & Computer Science, University of Massachusetts Amherst. I am advised by Prof. Negin Rahimi. My broad research interests are primarily in Information Retrieval (IR) and Natural Language Processing (NLP). Prior to this, I received a Dual Degree (Bachelors + Masters) in Mathematics and Computing from IIT Kharagpur, India where I worked on the paradigm of tool-augmentation for mathematical reasoning in LLMs, supervised by Prof. Somak Aditya.

My overarching research goals are focused on the effective utilization of retrieval-based systems and external structured knowledge to improve multiple reasoning dimensions in LLMs. Currently, I am interested in developing advanced retrieval systems for reasoning-intensive tasks, where retrievers perform multi-step, complex reasoning across multiple turns, leveraging both intermediate outputs and previously retrieved information. I am working on agentic retrieval frameworks, using reinforcement learning to train retrievers with step-wise feedback in dynamic environments. My other research interests revolve around the use of neuro-symbolic methods in NLP, and trying to make LLMs safer to use in social contexts.


News Slider

Updates/News

(08/25) RaDeR: Reasoning-aware Dense Retrieval Models accepted as Main Conference Paper at EMNLP 2025, Suzhao, China!

(09/24) Started my PHD at UMass Amherst - advised by Prof. Negin Rahimi!

(06/24) Presented my main conference paper on "MATHSENSEI: A Tool-Augmented Large Language Model for Mathematical Reasoning" at NAACL 2024, Mexico City.

(12/23) Completed my internship at Rakuten Global Inc., Language and Speech Team, RIT India.

Publications

Paper 1 Figure

RaDeR: Reasoning-aware Dense Retrieval Models

Debrup Das, Sam O' Nuallain, Razieh Rahimi
Main Conference Paper at EMNLP 2025, Suzhao, China

Paper 2 Figure

MATHSENSEI: A Tool-Augmented Large Language Model for Mathematical Reasoning

Debrup Das, Debopriyo Banerjee, Somak Aditya, Ashish Kulkarni
Main Conference Paper at NAACL 2024, Mexico City

Paper 3 Figure

SMAB: MAB based word Sensitivity Estimation Framework and its Applications in Adversarial Text Generation

Saurabh Kumar Pandey, Sachin Vashistha, Debrup Das, Somak Aditya, Monojit Choudhury
Main Conference Paper at NAACL 2025, New Mexico