Inkit Padhi

Member of Technical Staff at IBM Research, New York

ML & NLP Researcher

Hi! My name is Inkit. I’m a researcher interested in fields of Machine Learning (ML) and Natural Language Processing (NLP). My current work centers around improving the safety, reliability, and trustworthiness of Large Language Models (LLMs).

Currently, my primary focus and interests lie in safety and alignment, leveraging synthetic data, developing effective steering strategies, and unlocking advanced reasoning abilities in LLMs.

My past research has encompassed various areas of deep learning, including but not limited to, learning representations for diverse modalities, counterfactual generation, interpretability, text style transfer, unsupervised learning, influence-based attribution and more. I began my research journey at USC/ISI under the guidance of Kevin Knight; our work on probing in sequence models established a foundational contribution in the field of interpretability.

Email $first_name.$last_name@gmail.com

Oct 2025 award

Best Paper Award at AIES 2025

"When in Doubt, Cascade: Towards Building Efficient and Capable Guardrails" received the Best Paper Award at AIES 2025.

Announcement →

Sep 2025 release

Granite Guardian 3.3 released

Granite Guardian 3.3 is here, featuring hybrid thinking mode and a 3rd-place finish on the LLM-AggreFact benchmark.

Model →

Apr 2025 talk

Presenting CAST at ICLR 2025

Presenting CAST on conditional activation steering at ICLR in Singapore.

Poster →

Apr 2025 award

Granite Guardian tops GuardBench leaderboard

Granite Guardian tops GuardBench, a leaderboard for guardrail models.

Blog →

Feb 2025 paper

Granite Guardian accepted at NAACL 2025 (Oral)

Granite Guardian has been accepted to the NAACL '25 Industry Track as an Oral presentation. The latest Granite Guardian 3.1 model now also covers agentic risks.

arXiv →

View all news

When in Doubt, Cascade: Towards Building Efficient and Capable Guardrails

Manish Nagireddy, Inkit Padhi, Soumya Ghosh, Prasanna Sattigeri

Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 2025

Best Paper

PDF DOI

@inproceedings{nagireddy2025cascade-guardrails,
  title     = {When in Doubt, Cascade: Towards Building Efficient and Capable Guardrails},
  author    = {Manish Nagireddy and Inkit Padhi and Soumya Ghosh and Prasanna Sattigeri},
  year      = {2025},
  booktitle = {Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society},
  doi       = {https://ojs.aaai.org/index.php/AIES/article/view/36676}
}

Granite Guardian: Comprehensive LLM Safeguarding

Inkit Padhi, Manish Nagireddy, Giandomenico Cornacchia, Subhajit Chaudhury, Tejaswini Pedapati, Pierre Dognin, Keerthiram Murugesan, Erik Miehling, Martin Santillan Cooper, Kieran Fraser, Giulio Zizzo, Muhammad Zaid Hameed, Mark Purcell, Michael Desmond, Qian Pan, Inge Vejsbjerg, Elizabeth M. Daly, Michael Hind, Werner Geyer, Ambrish Rawat, Kush R. Varshney, Prasanna Sattigeri

Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Industry Track), 2025

arXiv PDF DOI

@inproceedings{padhi2025graniteguardian,
  title     = {Granite Guardian: Comprehensive LLM Safeguarding},
  author    = {Inkit Padhi and Manish Nagireddy and Giandomenico Cornacchia and Subhajit Chaudhury and Tejaswini Pedapati and Pierre Dognin and Keerthiram Murugesan and Erik Miehling and Martin Santillan Cooper and Kieran Fraser and Giulio Zizzo and Muhammad Zaid Hameed and Mark Purcell and Michael Desmond and Qian Pan and Inge Vejsbjerg and Elizabeth M. Daly and Michael Hind and Werner Geyer and Ambrish Rawat and Kush R. Varshney and Prasanna Sattigeri},
  year      = {2025},
  booktitle = {Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Industry Track)},
  doi       = {10.18653/v1/2025.naacl-industry.49}
}

Programming Refusal with Conditional Activation Steering

Bruce W. Lee, Inkit Padhi, Karthikeyan Natesan Ramamurthy, Erik Miehling, Pierre Dognin, Manish Nagireddy, Amit Dhurandhar

International Conference on Learning Representations, 2025

Spotlight

arXiv PDF DOI

@inproceedings{lee2025programming-refusal,
  title     = {Programming Refusal with Conditional Activation Steering},
  author    = {Bruce W. Lee and Inkit Padhi and Karthikeyan Natesan Ramamurthy and Erik Miehling and Pierre Dognin and Manish Nagireddy and Amit Dhurandhar},
  year      = {2025},
  booktitle = {International Conference on Learning Representations},
  doi       = {https://openreview.net/forum?id=Oi47wc10sm}
}

Value Alignment from Unstructured Text

Inkit Padhi, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Manish Nagireddy, Pierre Dognin, Kush Varshney

Conference on Empirical Methods in Natural Language Processing (Industry Track), 2024

PDF

@inproceedings{padhi2024value-alignment,
  title     = {Value Alignment from Unstructured Text},
  author    = {Inkit Padhi and Karthikeyan Natesan Ramamurthy and Prasanna Sattigeri and Manish Nagireddy and Pierre Dognin and Kush Varshney},
  year      = {2024},
  booktitle = {Conference on Empirical Methods in Natural Language Processing (Industry Track)}
}

ComVas: Contextual Moral Values Alignment System

Inkit Padhi, Pierre Dognin, Jesus Rios, Ronny Luss, Swapnaja Achintalwar, Matthew Riemer, Miao Liu, Prasanna Sattigeri, Manish Nagireddy, Kush R. Varshney, Djallel Bouneffouf

International Joint Conference on Artificial Intelligence (Demo Track), 2024

Demo