I am a fourth year CS PhD student at EPFL working with Prof. Michael Kapralov. I am broadly interested in algorithms for massive datasets, with a focus on randomized numerical linear algebra, high-dimensional data analysis and optimization. Recently, I have been working on optimizing memory and runtime complexity of LLM inference and training. I spent the fall of 2024 as an Applied Science intern at Amazon Luxembourg.

Previously, I worked with Prof. Ola Svensson during my MSc on clustering problems. Before that I worked with Prof. Anirban Dasgupta and Prof. Dinesh Garg (IBM Research, Bengaluru) on randomized linear algebra. I also spent a summer at Caltech on a SURF fellowship working with Dr. Ashish Mahabal on deep learning for astronomy.

Publications

BalanceKV: KV Cache Compression through Discrepancy Theory
Insu Han, Michael Kapralov, Ekaterina Kochetkova, KS, Amir Zandieh.
[arxiv]

Improved Algorithms for Kernel Matrix-Vector Multiplication
Piotr Indyk, Michael Kapralov, KS, Tal Wagner.
ICLR 2025
"We design fast algorithms for processing attention matrices in long-context LLMs"
Best Paper at ICML 2024 workshop on Long Context Foundation Models.
[openreview]

Sublinear Time Low-Rank Approximation of Toeplitz Matrices
Cameron Musco and KS.
SODA 2024
[arxiv]

Toeplitz Low-Rank Approximation with Sublinear Query Complexity
Michael Kapralov, Hannah Lawrence, Mikhail Makarov, Cameron Musco, and KS.
SODA 2023
[arxiv]

Towards Non-Uniform k-Center with Constant types of Radii
Xinrui Jia, Lars Rohwedder, KS, Ola Svensson.
SOSA 2022
[arxiv]

Fair Colorful k-Center Clustering
Xinrui Jia, KS, Ola Svensson.
Math. Programming 2021
Preliminary version in IPCO 2020
[arxiv] [talk]

Improved linear embeddings via Lagrange duality
KS, Dinesh Garg, Anirban Dasgupta.
Machine Learning, 2019
[paper]

Deep-learnt classification of light curves
Ashish Mahabal, KS, Fabian Gieseke, Akshay Pai, S George Djorgovski, Andrew J Drake, Matthew J Graham.
IEEE SSCI 2017
[arxiv]

Publications

Teaching

Service