Saswat Das

Saswat Das

PhD Student in Computer Science

SEAS, University of Virginia

About Me

I am a PhD student at the Department of Computer Science of the University of Virginia (UVA). I am fortunate to be advised by Dr. Ferdinando Fioretto as a member of the RAISE Group at UVA. My research interests are broadly situated within the fascinating area of trustworthy/responsible AI; more precisely, these include differential privacy, contextual integrity, algorithmic fairness, adversarial robustness. I am also interested in cryptography and topics related to security and (agentic) LLMs.

Prior to this, I pursued an Integrated M.Sc. (BS+MS) degree at the National Institute of Science Education and Research (NISER) with a major in Mathematics and a minor in Computer Science. I also was a student researcher at the School of Computer Sciences at NISER (2021-2023).

Reach out to me anytime for discussing ideas/collaborations, regarding opportunities, and to give talks on my research. I’m open to research internship opportunities as well!

Interests
  • Differential Privacy
  • Privacy-Preserving and Fair Machine Learning
  • Trustworthy AI
  • Cryptography/Security
  • Agentic LLMs
Education
  • PhD (Computer Science), 2023 - Present

    SEAS, University of Virginia

  • Integrated M.Sc. (BS+MS) (Mathematics Major and Computer Science Minor), 2018 - 23

    National Institute of Science Education and Research, HBNI

Recent News

All news»

May 2025: New preprint on “Disclosure Audits for LLM Agents” out on arXiv!

May 2025: Starting a research internship this summer at the Pacific Northwest National Laboratory to work on data privacy and security!

May 2025: Google Privacy Seminar: Presented our work on “Disclosure Audits for LLM Agents” with Nando Fioretto

Mar 2025: TOC4Fairness Invited Talk: Presented our work on “Fairness Issues and Mitigations in (Differentially Private) Socio-demographic Data Processes” on Mar 5 with Joon Ko

Mar 2025: Best Paper Award at CoLoRAI Workshop at AAAI-25 awarded to our paper titled “Low-rank Finetuning for LLMs is Inherently Unfair” 🏆!

Experience

 
 
 
 
 
SEAS, University of Virginia
PhD Student
Aug 2023 – Present Charlottesville, VA, USA
 
 
 
 
 
Pacific Northwest National Laboratory
PhD Intern
Jun 2025 – Aug 2025 Charlottesville, VA, USA (Remote)
 
 
 
 
 
EECS, Syracuse University
Visiting Research Scholar/Collaborator
Jun 2022 – May 2023 Syracuse, NY, USA
 
 
 
 
 
SM Lab, NISER, HBNI
Student Researcher
Mar 2021 – May 2023 Odisha, IN
 
 
 
 
 
School of Computer Sciences, NISER, HBNI
Winter Intern
School of Computer Sciences, NISER, HBNI
Dec 2019 – Jan 2020 Odisha, IN
 
 
 
 
 
School of Computer Sciences, NISER, HBNI
Summer Intern
School of Computer Sciences, NISER, HBNI
May 2019 – Jul 2019 Odisha, IN

Articles, Book Chapters, and Publications

Quickly discover relevant content by filtering publications.
(2025). Disclosure Audits for LLM Agents. Under Review.

PDF Cite

(2024). Fairness Issues and Mitigations in (Differentially Private) Socio-demographic Data Processes. AAAI-25.

PDF Cite

(2024). Low-rank finetuning for LLMs: A fairness perspective. CoLoRAI @ AAAI-25.

PDF Cite

(2024). Disparate Impact on Group Accuracy of Linearization for Private Inference. ICML-24.

PDF Cite

(2023). Finding ε and δ of Statistical Disclosure Control Systems. AAAI-24.

PDF Cite Poster DOI

(2023). Advances in Differential Privacy and Differentially Private Machine Learning - A Survey. In Information Technology Security (Springer Nature).

DOI

(2022). Fair Context-Aware Privacy Threat Modelling. In WPTM 2022, USENIX.

PDF Cite

Contact