Memento: Note-Taking for Your Future Self, arXiv preprint, 2025
N²: A Unified Python Package and Test Bench for Nearest Neighbor-Based Matrix Completion, arXiv preprint, 2025
PhantomWiki: On-Demand Datasets for Reasoning and Retrieval Evaluation, ICML, 2025
Low-Rank Thinning, ICML, 2025
Supervised Kernel Thinning, NeurIPS, 2024
Research Experience
Currently working on using distribution compression to speed up training and inference of large-language models at Cornell.
Education
Third-year PhD student in Computer Science at Cornell, advised by Raaz Dwivedi and Kilian Q. Weinberger; Undergraduate at Yale, worked with Andre Wibisono, Zhong Shao, and Cormac O'Dea.
Background
Research interests: Using distribution compression (a.k.a. "thinning") to speed up training and inference of large language models (LLMs). Long-term goal: Enable LLM agents to efficiently search over large and dynamic datastores.
Miscellany
Contact: Email / Google Scholar / LinkedIn / Github