Subhendu Khatuya

About

I am a PhD candidate in Computer Science and Engineering at the Indian Institute of Technology, Kharagpur, and have submitted my doctoral thesis under the supervision of Dr. Pawan Goyal and Dr. Niloy Ganguly. My broader research interests lie in Natural Language Processing and Instruction tuning of LLMs and Reasoning with a particular emphasis on domain specific applications.

My doctoral research focuses on Instruction Tuning of LLMs for financial applications where I developed models that combine generative retrieval, metadata-aware label semantics, dynamic in-context learning, and hybrid extractive–abstractive pipelines to enable reliable reasoning under noisy evidence, rare labels, and long-context constraints. This work has led to publications in top-tier venues, including EMNLP 2025 , ACL 2025 , NAACL 2024 , SIGIR 2024 , ACL 2023 and IJCAI 2024 (doctoral consortium). Beyond my core thesis work, my research contributions also include publications at AAAI 2025 reflecting my broader interests in temporal modeling and human behavior understanding. I have also contributed to large-scale social media analysis, including an AAAI–ICWSM’24 publication that investigates how the COVID-19 reshaped anti-vaccine discourse on Twitter using instruction-tuned generative classifiers.
Prior to my Ph.D., I worked as a Data Scientist at Ernst & Young, where I contributed as a core member of the AI/ML team. Before that, I completed my M.S (by research) in Computer Science and Engineering at the Indian Institute of Technology Kharagpur, working under the supervision of Prof. Bivas Mitra and Prof. Niloy Ganguly. My master’s thesis explored novel ML algorithm for anomaly detection and graph based method for troubleshooting. Both of these works resulted in publications at top-tier venues, including INFOCOM 2018 and PAKDD 2018 .

Employment

• Ernst and Young Senior Associate
  2018 - 2022

Education

• Indian Institute of Technology Kharagpur PhD
  2022 - 2026

Projects & Funding

Projects & funding information is unavailable.

Publications (10)

• Program of Thoughts for Financial Reasoning: Leveraging Dynamic In-Context Examples and Generative Retrieval
  Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing 2025 DOI: 10.18653/v1/2025.emnlp-main.1577
  +Save
• Label-semantics Aware Generative Approach for Domain-Agnostic Multilabel Classification
  Findings of the Association for Computational Linguistics: ACL 2025 2025 DOI: 10.18653/v1/2025.findings-acl.1145
  +Save
• ExPERT: Modeling Human Behavior Under External Stimuli Aware Personalized MTPP
  Proceedings of the AAAI Conference on Artificial Intelligence 2025
  +Save
• Parameter-Efficient Instruction Tuning of Large Language Models For Extreme Financial Numeral Labelling
  Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) 2024 DOI: 10.18653/v1/2024.naacl-long.410
  +Save
• How COVID-19 has Impacted the Anti-Vaccine Discourse: A Large-Scale Twitter Study Spanning Pre-COVID and Post-COVID Era
  Proceedings of the International AAAI Conference on Web and Social Media 2024
  +Save
• Instruction-guided bullet point summarization of long financial earnings call transcripts
  Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval 2024
  +Save
• Financial Numeric Extreme Labelling: A dataset and benchmarking
  Findings of the Association for Computational Linguistics: ACL 2023 2023 DOI: 10.18653/v1/2023.findings-acl.219
  +Save
• Adele: Anomaly detection from event log empiricism
  IEEE INFOCOM 2018-IEEE Conference on computer communications 2018
  +Save
• GBTM: Graph Based Troubleshooting Method for Handling Customer Cases Using Storage System Log
  Pacific-Asia Conference on Knowledge Discovery and Data Mining 2018
  +Save
• Parameter efficient instruction tuning of LLMs for financial applications
  +Save