Welcome! My name is Jerry Li, and I am a first-year MSCS Student at USC. I'm currently investigating a variety of topics ranging from embedding spaces of foundational models to LLM robustness. My previous experiences primarily lie within autonomous driving; however, I've also worked on various projects involving applications and experiments with LLMs. I was grateful to be advised by Dr. Hang Qiu and Dr. Vagelis Papalexakis during my undergraduate studies at the University of California, Riverside (UCR).
Education
University of Southern California
M.S. in Computer Science
University of California, Riverside
B.S. in Computer Science
Publications
arxiv 2025
Beyond ROUGE: N-Gram Subspace Features for LLM Hallucination Detection
Jerry Li, Evangelos Papalexakis
Experience
Research Assistant — FORTIS Lab @ USC
Foundations Of Robust Trustworthy Intelligent Systems
Machine Learning Engineer Intern — NexaAI
Research Intern — WCSNG Lab @ UCSD
Wireless Communications Sensing and Networking
Research Assistant — CISL Lab @ UCR
Collaborative Intelligent Systems Lab
Portfolio
cuda-miniGPT
A CUDA implementation of the inference pass of a Decoder only transformer.
Differential Multimodal Transformer
For CS228 (Deep Learning), we implemented the Differential Attention proposed in the Differential Transformer Paper.
PaperSync - Berkeley LLM Hackathon
Berkeley LLM Hackathon Project where we made a X (Twitter) like platform but for Research only.
TicTacToe Neural Network
Trained a PyTorch Neural Network to learn how to play TicTacToe.
C++ Grad Engine
C++ based grad engine to replicate a mini PyTorch with CPU parallelization support.