Rishav Chourasia

Betterdata

PIXEL @ One-North

10 Central Exchange Green

Singapore 138649

I work on differential privacy — the mathematics of using people’s data without exposing the people. I’m currently the Head of Product at Betterdata, a Singapore deep-tech startup building privacy-preserving synthetic data, where I earlier joined as a privacy engineer and wrote the privacy-auditing engine at the heart of its platform.

I recently completed my Ph.D. at the NUS School of Computing, advised by Prof. Xiaokui Xiao. My research spans private optimization, privacy auditing, and machine unlearning. Along the way I proved the first \(O(1)\) privacy bound for Noisy Gradient Descent (a NeurIPS 2021 spotlight), and my team ran the first independent audit of Apple’s DifferentialPrivacy.framework — the code privatising analytics on billions of devices — which won a Distinguished Paper Award at IEEE S&P 2026.

Before Singapore, I earned my B.Tech in Computer Science from IIT Guwahati, spent a summer as a student researcher at Google Brain, and held a research fellowship at the Max Planck Institute for Software Systems.

Take a look at my publications, projects, and blog — or reach out via any of the links below.

news

May 19, 2026	Our paper Auditing Apple’s DifferentialPrivacy.framework won a Distinguished Paper Award at IEEE S&P 2026! We ran the first independent audit of the differential-privacy code running on billions of Apple devices.
Dec 01, 2025	I’ve taken on the role of Head of Product at Betterdata, leading our privacy-preserving synthetic-data platform.
Jun 10, 2025	My paper on One-Run Privacy Auditing for Structured Generative and Foundation Models was accepted at the FMSD workshop in ICML 2025.
May 15, 2025	I’ll be defending my Ph.D. Thesis on May 26th. Venue: SR9 COM1 02-09, School of Computing, National University of Singapore.
May 07, 2025	I’m giving a talk at FORC 2025, being organized this year at Stanford University, on June 4-7, 2025.

latest posts

Jun 23, 2025	Auditing ε-DP using Reconstruction Attacks
Feb 15, 2023	Failure in Machine Unlearning and How to Guarantee Deletion
Nov 22, 2021	DP Dynamics of Noisy Gradient Descent (Part 2)

selected publications

IEEE S&P 2026

Auditing Apple’s DifferentialPrivacy.framework: Implementation Bugs, Misconfigurations, and Practical Risks

Rishav Chourasia, Ergute Bao, Uzair Javaid, and Xiaokui Xiao

In 2026 IEEE Symposium on Security and Privacy (SP), 2026

Distinguished Paper Award

arXiv HTML DOI Bib

@inproceedings{chourasia2026auditing,
  title = {Auditing Apple's DifferentialPrivacy.framework: Implementation Bugs, Misconfigurations, and Practical Risks},
  author = {Chourasia, Rishav and Bao, Ergute and Javaid, Uzair and Xiao, Xiaokui},
  booktitle = {2026 IEEE Symposium on Security and Privacy (SP)},
  year = {2026},
  pages = {1815--1834},
  publisher = {IEEE},
  doi = {10.1109/SP63933.2026.00225},
}

FORC25, PPAI25

Laplace Transform Interpretation of Differential Privacy

Rishav Chourasia, Uzair Javaid, and Biplab Sikdar

In Proceedings of the 6th Symposium on Foundations of Responsible Computing (FORC), LIPIcs Vol. 329, 2025

arXiv HTML DOI Bib

@article{chourasia2024laplace,
  abbr2 = {& TPDP25},
  abbr3 = {(oral)},
  title = {Laplace Transform Interpretation of Differential Privacy},
  author = {Chourasia, Rishav and Javaid, Uzair and Sikdar, Biplab},
  journal = {In Proceedings of the 6th Symposium on Foundations of Responsible Computing (FORC), LIPIcs Vol. 329},
  pages = {11:1--11:23},
  publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f\"ur Informatik},
  doi = {10.4230/LIPIcs.FORC.2025.11},
  year = {2025}
}

ICML 2023

Forget Unlearning: Towards True Data-Deletion in Machine Learning

Rishav Chourasia and Neil Shah

In Proceedings of the 40th International Conference on Machine Learning, 2023

arXiv HTML Bib

@inproceedings{pmlr-v163-chourasia23forget,
  abbr2 = {& FORC 2023},
  title = {Forget Unlearning: Towards True Data-Deletion in Machine Learning},
  author = {Chourasia, Rishav and Shah, Neil},
  booktitle = {Proceedings of the 40th International Conference on Machine Learning},
  year = {2023},
  volume = {202},
  pages = {6028--6073},
  series = {Proceedings of Machine Learning Research},
  publisher = {PMLR},
}

NeurIPS 2021

Differential Privacy Dynamics of Langevin Diffusion and Noisy Gradient Descent

Rishav Chourasia, Jiayuan Ye, and Reza Shokri

In Advances in Neural Information Processing Systems, 2021

Spotlight

arXiv HTML Bib

@inproceedings{NIPS2021_7c6c1a7b,
  author = {Chourasia, Rishav and Ye, Jiayuan and Shokri, Reza},
  booktitle = {Advances in Neural Information Processing Systems},
  editor = {Ranzato, M. and Beygelzimer, A. and Dauphin, Y. and Liang, P.S. and Vaughan, J. Wortman},
  pages = {14771--14781},
  publisher = {Curran Associates, Inc.},
  title = {Differential Privacy Dynamics of Langevin Diffusion and Noisy Gradient Descent},
  volume = {34},
  year = {2021},
}