Towards Provably Privacy-Preserving AI in the Age of Foundation Models (recording), CSML Workshop @ IISc, Early Career Highlights @ CODS Conference November & December 2025.
Towards Provably Privacy-Preserving AI in the Age of Foundation Models, WSAI Faculty Talks @ IITM, September 2025.
Near-Optimal Private Learning with Correlated Noise Mechanisms (slides, recording), SCTS Seminar at TIFR, July 2025.
InvisibleInk: High-Utility and Low-Cost Text Generation with Differential Privacy. Microsoft Research, June 2025.
Towards User-Level Differential Privacy at Scale (slides), IISc, Microsoft Research India Lab, Amazon Tech Talks, February & March 2025
Was My Data Used to Train a LLM? (slides), National Symposium on AI in Healthcare (IIT Jodhpur), November 2024.
Was My Data Used to Train a LLM? (slides), GenAI in Cybersecurity Symposium (MeitY/DSCI), October 2024.
Robust Aggregation for Federated Learning (slides, recording), IEEE Webinar, June 2024.
Towards User-Level Differential Privacy at Scale (slides), IIT Hyderabad, April 2024.
Correlated Noise Provably Beats Independent Noise for Differentially Private Learning (slides), Laboratoire Jean Kuntzmann, UGA (Grenoble, France), February 2024.
Towards User-Level Differential Privacy at Scale (slides), Google Federated Learning Seminar (Postdoc final presentation), February 2024.
Unleashing the Power of Randomization in Auditing Differentially Private ML, GraPFiCs Workshop at UC Santa Cruz, October 2023.
MAUVE Scores for Generative Models, ETH Zürich (Zurich, Switzerland), June 2023.
Distributionally Robust Federated Learning: Differential Privacy and Fast Optimization, EPFL (Lausanne, Switzerland), June 2023.
Distributionally Robust Federated Learning with Differential Privacy (slides), SIAM Conference on Optimization (OP23) (Seattle, WA, USA), May 2023.
Towards Next-Generation ML/AI: Robustness, Optimization, Privacy (slides), Faculty Job Talk (Multiple locations), January 2023.
Tackling Distribution Shifts in Federated Learning with Superquantile Aggregation (slides), NeurIPS DistShift (Spotlight Talk) (New Orleans, LA, USA), December 2022.
Federated Learning with Heterogeneous Data: A Superquantile Optimization Approach (slides), International Conference on Continuous Optimization (ICCOPT) (Bethlehem, PA, USA), July 2022.
Federated Learning with Partial Model Personalization (slides), ICML (Spotlight Talk) (Baltimore, MD, USA), July 2022.
Federated Learning: Heterogeneity, Robustness, and Optimization. (slides), PhD Defense (University of Washington, Seattle, WA, USA), June 2022.
MAUVE: Measuring the Gap Between Neural Text and Human Text (slides), Microsoft Research Asia, March 2022.
Federated Learning with Heterogeneous Data: A Superquantile Optimization Approach (slides), INFORMS Optimization Society Conference (Greenville, SC, USA), March 2022.
MAUVE: Measuring the Gap Between Neural Text and Human Text (slides), Stanford NLP Seminar, March 2022.
MAUVE: Measuring the Gap Between Neural Text and Human Text (slides), NeurIPS Oral Presentation, December 2021.
Statistics of Evaluating Generative Models with Divergence Frontier (slides), FAIR Science of Deep Learning, October 2021.
A Superquantile Approach to Federated Learning with Heterogeneous Devices, IFDS Ethics and Algorithms, September 2021.
Robust Aggregation for Federated Learning, FL-ICML Long Presentation, July 2020.
Robust Aggregation for Federated Learning, Federated Learning One World Seminar, July 2020.
A Smoother Way to Train Structured Prediction Models, Facebook AI Research, July 2019.
