My Master’s thesis is on risk of Alzheimer’s Disease prediction using resource-efficient and explainable LLMs, mentored by Dr. Jennifer Dworak, Dr. Eric Larson and Dr. Jia Zhang. This thesis aims to develop a predictive metric for risk of Alzheimer’s Disease based on natural language, using LLM finetuning, embedding analysis, and other ML and NLP techniques.

I have been developing a regression algorithm with LLMs to predict personality scores under Dr. Mehak Gupta, involving research in deep learning and NLP with applications in healthcare. Our papers, AI Assistant for Socioeconomic Empowerment Using Federated Learning has been published to NLP4DH at NAACL 2025; Language-Based AI Modeling of Personality Traits and Pathology from Life Narrative Interviews has been published to Journal of Psychopathology and Clinical Science

My independent research on LLMs as recommender systems mentored by Dr. Michael Hahsler of Computer Science at the Lyle School of Engineering, and Dr. Peter Moore of Mathematics at the Dedman College of Humanities and Sciences. A Recommender System Architecture for University Curriculum Advising has been published to AAAI Spring Symposium and to NCUR. The research was funded by the Robert Mayer Undergraduate Research Fellowship.

This summer, I trained AlphaFold based Stable Diffusion models that predict multi-frame protein trajectory in Dr. Peng Tao’s Proteins and Computers Lab. I also developed a Python package, PyTAO, to assist ONIOM calculations in AMBER and Gaussian.

In Dr. Corey Clark’s HuMIn Game Lab, I am developing a Graph-RAG vector/graph database using MCP, striving for faster and more relevant text data retrieval and lookup.

• Machine Learning:

Transformers, Stable Diffusion, Graph-RAG, NLP

• ML for health:

LLM for Text Embedding, Knowledge Representation, Clinical Diagnostic Metric

• Human-Compatible AI:

Human Centered Computing, Recommender Systems, Human in the Loop

• Explainable AI (XAI):

Data Visualization, Embedding Analysis

Development Experience

• PyTAO, a Python package to assist ONIOM calculations.

• Social Panacea admin portal: a web application for university student social life, using Python, fastAPI, firebase and ReactJS.

• Automative Debug Assistant Job System an LLM powered job system for automotive debugging, written in C++ and javascript. Utilized mutex locks for concurrency control, designed a custom programming language for job management.

• AI Club President, organizing industry expert forum and workshops for AI enthusiasts at SMU.