about me
I am an assistant professor at the Department of Computer Science and Engineering, University of North Texas. I have two broad research interests: 1. Explaining the behavior and limitations of LLMs with a focus on model bias and reasoning abilities; and 2. Scholarly information processing to enhance users’ experience with digital libraries. Some example tasks are a) extracting information from scientific figures, b) understanding reproducibility of scientific articiles, and automatically generating limitations of scientific papers.
The best way to reach me is by email: sagnikrayc at gmail dot com. I am not on any social media (X/bluesky/Insta/FB). The linkedin profile is not maintained.
News
- JCDL 2025: I am a co-PC chair for the 2025 ACM/IEEE Joint Conference of Digital Libraries (JCDL’25). This time JCDL is completely virtual and synchronous, send your papers! Full CFP here: https://2025.jcdl.org/call-for-papers/
- Pre-print: On measuring political bias in LLMs (or the futility of it): A Detailed Factor Analysis for the Political Compass Test: Navigating Ideologies of Large Language Models
- EMNLP 2025: Findings: BAGELS: Benchmarking the Automated Generation and Extraction of Limitations from Scholarly Text. The camera-ready would be uploaded on arxiv soon, with dataset and code.
- SDP@ACL 2025: Predicting The Scholarly Impact of Research Papers Using Retrieval-Augmented LLMs
- EMNLP 2024: Paper on generalization awarded prize - GenBench@EMNLP2024
- August 2024: Joined the University of North Texas as an assitant professor of computer science.
- July 2024: Paper accepted at CIKM 2024 on reproducibility – more information + arxiv link coming soon!
- Dec 2023: EMNLP travel + talk.
- Nov 2023: PLOS ONE paper on gender bias in LLMs
- Oct 2023: Two papers accepted: 1. ConLL paper on Edge probing and 2. EMNLP main on interaction explanations.
Experience
Previously, I was an NLP Scientist at the National Board of Medical Examiners. Before that, I worked at the University of Michigan Medical School on information extraction models on clinical text and their generalization abilities + models for early dementia prediction. I came to Michigan from the Department of Computer Science, UCPH, CopeNLU group, where I worked on the explainability of DNN models used in multi-hop reasoning systems, such as question answering, fact-checking and natural language inference.
During my Ph.D. from Penn State I worked in the CiteSeerX group on information extraction from scholarly figures and tables, information retrieval and crawling. Post Ph.D, I worked as an NLP/ML engineer at Interactions. I developed DNN models for large scale entity extraction and linking, dialog systems, and sentiment classification. I also contributed to a DNN library that was used as the ML backend for the company.