I'm currently working on visual information retrieval with Dr. Abhinav Shrivastava. Previously, I worked on building trustworthy LLMs via data attribution with Dr. Prathosh A.P. at IISc Bangalore, and on test-time alignment of LLMs at Purdue with Dr. Berkay Celik. I've also had a chance to work with Dr. Pengtao Xie at UCSD to predict histone signatures of non-coding RNAs. Overall, my current interests lie in the geometry of deep learning.

I graduated in Physics from IIT Delhi (2024) with a bachelor's thesis in computational neuroscience.

Contact nishant.sharma.iitd@gmail.com

News

  • Apr 2026 🇧🇷 Visiting ICLR 2026 in Rio de Janeiro.
  • Jan 2026 🎉 One paper accepted at ICLR 2026.
  • Jan 2026 🚀 Joining Stealth to work on visual information retrieval.
  • Oct 2025 🎉 Two papers accepted at NeurIPS workshops — FM4LS and Reliable ML from Unreliable Data.

Selected Publications

Authors who equally contributed to a publication are marked with a † (dagger).

  1. Kumar Shubham, Nishant Sharma, Karn Tiwari, Prathosh A.P.. Enhancing Trustworthiness of Fine-Tuned LLMs via Regularized Subset Selection. ICLR (2026). ICLR 2026PDFOpenReviewCode
  2. Nishant Sharma, Mohammad Atif Quamar, Pengtao Xie. Decoding Histone Modification Signatures of Non-Coding RNAs via Foundation Models. NeurIPS Workshop on Multi-modal Foundation Models and LLMs for Life Sciences (2025). NeurIPS--W 2025PDF
  3. Mohammad Atif Quamar, Mohammad Areeb, Nishant Sharma, Ananth Shreekumar, Jonathan Rosenthal, Muslum Ozgur Ozmen, Mikhail Kuznetsov, Z. Berkay Celik. Adaptive Blockwise Search: Inference-Time Alignment for Large Language Models. Preprint (2025). arXiv 2025arXivHTML
  4. Kumar Shubham, Nishant Sharma, Karn Tiwari, Prathosh A.P.. Trust, But Attribute: Tracing Impact of Data on Trustworthiness in Supervised LLM Fine-Tuning. NeurIPS Workshop on Reliable ML from Unreliable Data (2025). NeurIPS--W 2025PDFOpenReview

Projects

Writing / TIL