Research Overview
My research roughly aims to make powerful deep learning models more practical for use. My main focus is on the algorithmic side of LLM inference efficiency by leveraging inherent structures and patterns within the architecture, data, and/or pretrained weights.
I also devote a significant amount of time investigating how LLMs and diffusion models can be applied to challenging science problems, particularly in materials science, where there may be physical constraints and low error tolerance.
Previously, I have also worked on optimization.
Projects
Deep Learning Efficiency: Efficient algorithms/architectures to reduce inference costs, with a focus on LLMs.
Machine Learning in Materials Science: Scalable methods for materials data that have underlying physics relationships.
Previous Projects
Tensor Robust Principal Component Analysis: Low rank tensor recovery from sparsely corrupted data with theoretical guarantees and empirical analysis.
Traffic Routing: Optimal/stochastic vehicle routing in traffic networks.
Deep Learning for Neuroscience: Neuronal ion conductances prediction from voltage signals.
Awards
Wei Shen and Xuehong Zhang Presidential Fellowship (2024)
Liang Ji-Dian Graduate Fellowship (2023)
Michel and Kathy Doreau Graduate Fellowship in Electrical and Computer Engineering (2023)
NSF GRFP Honorable Mention (2023)
UC Berkeley High Distinction (2021)
|