Saptarshi Sengupta
PhD Student in Informatics
I'm a PhD student at The Pennsylvania State University working on domain-specific applications of LLMs, under the guidance of my advisor, Dr. Suhang Wang. Outside of work, I am an advocate for animal rights 🦙, enjoy reading 📖 (currently on an Alan Turing biography), cooking 🍲 and playing the guitar 🎸.
I'm actively seeking full-time positions as an applied research scientist/engineer. If you believe I might be a good fit, do reach out & I'd be happy to talk 😀
Research Interests
My research interests span various aspects of NLP such as QA, RAG, IR/Search, LLM-agents, model interpretability and low-resource languages. Overall, I'm interested in applying language technologies to challenging edge-cases which have either too much/little data. Through my work, I aim to develop methods for tackling real-world problems that are easy to use and cost-effective.
Publications
Pre-Prints
BioMol-MQA: A Multi-Modal Question Answering Dataset For LLM Reasoning Over Bio-Molecular Interactions
Saptarshi Sengupta, Shuhua Yang, Paul Kwong Yu, Fali Wang, Suhang Wang
ToolDreamer: Instilling LLM Reasoning Into Tool Retrievers
Saptarshi Sengupta, Zhengyu Zhou, Jun Araki, Xingbo Wang, Bingqing Wang, Suhang Wang, Zhe Feng
MAG-V: A Multi-Agent Framework for Synthetic Data Generation and Verification
Saptarshi Sengupta, Harsh Vashistha, Kristal Curtis, Akshay Mallipeddi, Abhinav Mathur, Joseph Ross, Liang Gou
Published Work
TOP-Training: Target-Oriented Pretraining for Medical Extractive Question Answering
Saptarshi Sengupta, Connor Heaton, Shreya Ghosh, Wenpeng Yin, Preslav Nakov, Suhang Wang
International Conference on Computational Linguistics (COLING), 2025
Exploring Language Model Generalization in Low-Resource Extractive QA
Saptarshi Sengupta Wenpeng Yin, Preslav Nakov, Shreya Ghosh, Suhang Wang
International Conference on Computational Linguistics (COLING), 2025
Towards Efficient Methods in Medical Question Answering using Knowledge Graph Embeddings
Saptarshi Sengupta Connor Heaton, Suhan Cui, Soumalya Sarkar, Prasenjit Mitra
IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2024
Improving Semantic Similarity with Cross-Lingual Resources: A Study in Bangla—A Low Resourced Language
Rajat Pandit, Saptarshi Sengupta, Sudip Kumar Naskar, Niladri Sekhar Dash, Mohini Mohan Sardar
Informatics journal, 2019
Word sense induction in bengali using parallel corpora and distributional semantics
Saptarshi Sengupta Rajat Pandit, Parag Mitra, Sudip Kumar Naskar, Mohini Mohan Sardar
Journal of Intelligent & Fuzzy Systems: Applications in Engineering and Technology, 2019
Writing
Feed-Forward Neural Network From Scratch
I've always wanted to implement a simple FFNN from scratch just to see how the math works and really understand things at a deeper level. This is my attempt at creating something from an educational perspective, breaking down all the math in bits to be more accessible. Note: All of the code works but some final illustrations remain.
Experience
NLP and Large Language Model Intern
Robert Bosch LLC | May 2025 - August 2025
Performed research in tool retrieval for LLM-agents when dealing with a large number of tools. Proposed a new framework (ToolDreamer) for the same.
Machine Learning Applied Scientist Intern
Splunk | May 2024 - November 2024
Worked on synthetic data generation and LLM-agent trajectory verification for an internal AI assistant. Developed systems were implemented using the Autogen library