- LevAttention: Time, Space, and Streaming Efficient Algorithm for Heavy Attentions (with Ravindran Kannan, Chiranjib Bhattacharyya and David P. Woodruff)
- Differentially Private Vertical Federated Learning Primitives (with Vincent Cohen-Addad, Vahab Mirrokni and Peilin Zhong)
- Approximating the Top Eigenvector in Random Order Streams (with David P. Woodruff, NeurIPS 2024, selected as a spotlight)
- Faster Algorithms for Schatten-p Low Rank Approximation (with David P. Woodruff, RANDOM 2024)
- PolySketchFormer: Fast Transformers via Sketching Polynomial Kernels (with Vahab Mirrokni and Peilin Zhong, ICML 2024)
- High-Dimensional Geometric Streaming for Almost Low Rank Data (with Hossein Esfandiari, Vahab Mirrokni, David P. Woodruff and Peilin Zhong, ICML 2024)
- Optimal Communication Bounds for Classic Functions in Coordinator model and Beyond (with Hossein Esfanidari, Vahab Mirrokni, David P. Woodruff and Peilin Zhong, STOC 2024)
- Lower Bounds on Adaptive Sensing for Matrix Recovery (with David P. Woodruff, NeurIPS 2023)
- Pseudorandom Hashing for Space-bounded Computation with Applications in Streaming (with Rasmus Pagh, Mikkel Thorup and David P. Woodruff, FOCS 2023)
- Subquadratic Algorithms for Kernel Matrices via Kernel Density Estimation (with Ainesh Bakshi, Piotr Indyk, Sandeep Silwal and Samson Zhou, ICLR 2023)
- Sketching Algorithms and Lower Bounds for Ridge Regression (with David P. Woodruff, ICML 2022)
- Near-Optimal Algorithms for Linear Algebra in the Current Matrix Multiplication Time (with Nadiia Chepurko, Kenneth L. Clarkson and David P. Woodruff, SODA 2022)
- Reduced-Rank Regression with Operator Norm Error (with David P. Woodruff, COLT 2021)
- Dimensionality Reduction for Sum-of-Distances Metric (with Zhili Feng and David P. Woodruff, ICML 2021, selected for long talk)
- Robust k-means++ (with Amit Deshpande and Rameshwar Pratap, UAI 2020)
- Optimal Deterministic Coresets for Ridge Regression (with David P. Woodruff, AISTATS 2020)
Research Experience
Researcher in the Algorithms and Optimization team at Google.
Education
PhD: Carnegie Mellon University, Advisor: Prof. David P. Woodruff; Undergraduate: IIT Delhi.
Background
Research interests: algorithms, particularly randomized algorithms for Numerical Linear Algebra and matrix methods in Big Data settings. Professional field: Algorithms and Optimization.
Miscellany
Email: p followed by last name at google dot com or me at first name last name dot com