Hi, I am Amrita Roy Chowdhury*. I am an assistant professor in the CS department at the University of Michigan, Ann Arbor.

I analyze, design and build systems that enable safe data analytics in a decentralized setting without ever "seeing'' the data. My research empower individuals with full-stack data privacy solutions that deliver functionally-rich systems that are not only provably private but also tailored to meet application-specific constraints, ensuring their seamless deployment in the real-world. I do this via principled exploration of the synergy between differential privacy and cryptography, exposing their rich and multi-faceted interconnections in both theory and practice.

Here's a link to my CV (Updated Jan '25).

I am actively looking for students, please reach out if you are interested.

Recent News

Aug 25: I will be attending the Secure Computation workshop at the Simons Institute.

Jul 25: Our work Prεεmpt on sanitizing prompts for LLMs will be presented at NDSS '26.

May 25: Our paper on robust graph analysis will be presented at ASIACCS '25.

Apr 25: I will be attending the Governance Mechanism workshop.

Mar 25: I am giving an invited talk at the e-HAIL seminar.

Feb 25: I will be serving as a PC member for IEEE S&P '26.

Jan 25: I will be giving a talk at IIT-Madras.

Jan 25: I will be giving an invited talk at the Bengaluru Crypto Day at IISc Bengaluru.

Dec 24: Our paper on refuting membership inference attacks will be presented at IEEE SaTML '25.

Dec 24: I will be serving as a PC member for CCS '25.

Nov 24: I am giving an invited talk at the ML4Crypto workshop.

Oct 24: Check out my interview by the EnCORE magazine.

Sept 24: I am giving a keynote talk at the AACD workshop @CCS.

Aug 24: Our paper on metric DP at the user-level will be presented at CCS '24.

Aug 24: I will be serving as a PC member for USENIX Security '25, SaTML '25 and as a senior reviewer (area chair) for WWW '25.

Aug 24: I will be giving a keynote talk at TPDP.

May 24: FairProof won the best paper award at the Private ML@ICLR'24 workshop.

May 24: Our work FairProof on confidential and certifiable fairness of ML models will be presented at ICML '24.

Old News

Dec 23: Our work on prompt sanitization for LLMs will be presented at PPAI '24.

Nov 23: I will be giving a talk at the security seminar at UPenn.

Nov 23: I will be giving a talk at the Naval Postgraduate School.

Oct 23: I will be serving on the program committee for SaTML'24 and EuroS&P'24.

Oct 23: I will be giving a talk at the EnCORE Institute student social.

Sept 23: Check out our white paper on the security risks of GenAI based on the discussions from our workshop!

Sept 23: I will be a panelist for the Rising Stars in Data Science information session.

Aug 23: I will be an Area Chair for WebConf'24.

July 23: I will be giving a talk at the PriSec-ML seminar.

June 23: I helped organize the GenAI Risks Workshop. Stay tuned for some exciting updates!

June 23: I will be serving on the program committee for TPDP'23.

May 23: I will be giving a talk at the UCSD security seminar.

April 23: I will be giving a talk at Google.

March 23: I will be giving a talk at the FLOW seminar.

Feb 23: I won the award for the best poster at ITA'23.

Feb 23: I will be giving a talk at the Naval Applications of Machine Learning workshop.

Jan 23: I will be serving on the program committee for FORC '23, CCS '23 and FAccT '23.

Dec 22: I will be giving a talk at Cisco Research.

Nov 22: ShadowNet will be appearing at IEEE S&P 2023. Check out our talk!

Oct 22: I will be giving a talk at Amazon.

Oct 22: I will be giving a talk at Tufts University.

Sept 22: I will be co-organizing the PPML workshop this year co-located with FOCS. See you there!

Aug 22: EIFFeL will be appearing at CCS, 2022. Check this video for a 2-min summary!

Aug 22: I will be giving a talk at UC Berkeley AI4ALL.

July 22: Our work on strengthening order preserving encryption with differential privacy will be appearing at CCS, 2022.

May 22: I will be serving on the program committee for the first ever IEEE Conference on Secure and Trustworthy Machine Learning (IEEE SatML, 2023). Submit all your best work!

Apr 22: I will be serving on the program committee for ICDE, 2023.

March 22: I will be serving on the program committee for IEEE S&P, 2023 and VLDB, 2023.

Jan 22: Our work on the privacy implications of shuffling will be appearing at ICLR, 2022.

Dec 21: I will be serving on the program committee for CCS, 2022.

Nov 21: I have been selected for UChicago Rising Stars in Data Science.

Oct 21: I have been selected for Rising Stars in EECS, MIT.

Oct 21: I will be serving on the program committee for PPAI, 2022.

July 21: I have been selected as a CRA/CCC Computing Innovation Fellow, 2021. I will be working with Prof. Kamalika Chaudhuri at UCSD.

May 21: I will be serving on the program committee for USENIX Security, 2022.

May 21: I will be serving on the program committee for TPDP. Please submit all your awesome work!

Apr. 21: Honored to be selected as a finalist for the Facebook Fellowship, 2021 (top 3.5% out of over 2000 students worldwide).

Apr. 21: I will be interning at Facebook AI Research, NYC under Laurens Van der Maaten.

Feb. 21: Our work on eTAP, a privacy-enhancing trigger-action-platform, will be appearing at IEEE S&P, 2021.

Oct. 20: Selected for Rising Stars in EECS, UC Berkeley.

Oct. 20: I will be giving a talk to the Cryptography Group at MSR, Redmond.

Sept. 20: Our work on Kalεido, a system for real-time privacy-preserving processing of eye-tracking data, will be appearing at USENIX Security, 2021.

Preprints

What Really is a Member? Discrediting Membership Inference via Poisoning Neal Mangaokar, Ashish Hooda, Zhuohang Li, Bradley A. Malin, Kassem Fawaz, Somesh Jha, Atul Prakash, Amrita Roy Chowdhury
Differentially Private Quantiles with Smaller Error, Jacob Imola, Fabrizio Boninsegna, Hannah Keller, Anders Aamand, Amrita Roy Chowdhury, Rasmus Pagh
Vεrity: Verifiable Local Differential Privacy, James Bell-Clark, Adria Gascon, Baiyu Li, Mariana Raykova, Amrita Roy Chowdhury

Publications

SQUiD: Synthesizing Relational Databases from Unstructured Text, Mushtari Sadia, Zhenning Yang, Yunming Xiao, Ang Chen, Amrita Roy Chowdhury, EMNLP 2025
Prεεmpt: Sanitizing Sensitive Prompts for LLMs, Amrita Roy Chowdhury, David Glukhov, Divyam Anshumaan, Prasad Chalasani, Nicolas Papernot, Somesh Jha, Mihir Bellare, NDSS 2026
Robustness of Locally Differentially Private Graph Analysis Against Poisoning, Amrita Roy Chowdhury*, Jacob Imola*, Kamalika Chaudhuri, *Equal Contribution, ASIACCS 2025
On the Reliability of Membership Inference Attacks, Amrita Roy Chowdhury*, Zhifeng Kong*, Kamalika Chaudhuri, *Equal Contribution, IEEE SaTML 2025
Metric Differential Privacy at the User-Level, Jacob Imola, Amrita Roy Chowdhury, Kamalika Chaudhuri, CCS 2024
FairProof: Confidential and Certifiable Fairness for Neural Networks, Chhavi Yadav, Amrita Roy Chowdhury, Dan Boneh, Kamalika Chaudhuri, ICML 2024, Best Paper at Privacy-ICLR '24
ShadowNet: A Secure and Efficient System for On-device Model Inference, Zhichuang Sun, Ruimin Sun, Changming Liu, Amrita Roy Chowdhury, Somesh Jha, Long Lu, IEEE S&P 2023, Talk, Code
EIFFeL: Ensuring Integrity for Federated Learning, Amrita Roy Chowdhury, Chuan Guo, Somesh Jha, Laurens van der Maaten, CCS 2022, Best Poster at ITA '23
Strengthening Order Preserving Encryption with Differential Privacy, Amrita Roy Chowdhury, Bolin Ding, Somesh Jha, Weiran Liu, Jingren Zhou, CCS 2022
Privacy Implications of Shuffling, Casey Meehan, Amrita Roy Chowdhury, Kamalika Chaudhuri, Somesh Jha, ICLR 2022, Code
Data Privacy in Trigger-Action IoT Systems, Yunang Chen, Amrita Roy Chowdhury, Ruizhe Wang, Andrei Sabelfeld, Rahul Chatterjee, Earlence Fernandes, IEEE S&P 2021, Code
Kalεido: Real-Time Privacy Control for Eye-Tracking Systems, Jingjie Li, Amrita Roy Chowdhury, Younghyun Kim, Kassem Fawaz, USENIX Security 2021, Media Coverage
Cryptε: Crypto-Assisted Differential Privacy on Untrusted Servers, Amrita Roy Chowdhury, Chenghong Wang, Xi He, Ashwin Machanavajjhala, Somesh Jha, SIGMOD 2020, Code
Data-Dependent Differentially Private Parameter Learning for Directed Graphical Models, Amrita Roy Chowdhury, Theodoros Rekatsinas, Somesh Jha, ICML 2020
Concise Explanations of Neural Networks using Adversarial Training, Prasad Chalasani, Jiefeng Chen, Amrita Roy Chowdhury, Somesh Jha, Xi Wu, ICML 2020
Prεεch: A System for Privacy-Preserving Speech Transcription, Shimaa Ahmed, Amrita Roy Chowdhury, Kassem Fawaz, Parmeswaran Ramanathan, USENIX Security 2020, Demo

Students

Ph.D.

Mushtari Sadia (co-advised with Ang Chen)
Yiyi Sun
Samanway Sadhu

Miscellaneous

5 completely unsolicited facts about me --

If Ι were to describe myself and my work in a single sentence, I would say "I am indistinguishable from an anthropomorphic squirrel, who has humbugged her way into the Bluε Stockings Sociεty, for a PPT adversary."
I have an inordinate and perfervid predilection for circumlocutory and highfalutin verbiage. QED.
I would be well-disposed to having a dispassionate disquisition on whether to pet a Portuguese man o' war or a Porpita porpita at 3:07 am in the morning.
I would feature in the Forbes Billionaires List if the unit of currency were Scoville heat units consumed.
I would tell you all about my thoughts on "Why we should care for the Encephalartos woodii?", but I'm afraid I have exhausted my privacy budget.

Contact

If you would like to know more about my work OR challenge me to a tongue-twister battle, then drop me an email at aroyc@umich.edu.

P.S. My previous email address @ucsd is no longer active, so if you've emailed me and didn't receive a response, I apologize. Please use my @umich email address as above.

*My correct surname is 'Roy Chowdhury'.