About
Huirong Chai
Huirong Chai is a PhD researcher at Georgia State University working on human–computer interaction. Her work runs along two lines: measuring what language models actually cost at runtime on phones and edge hardware, and examining how the field evaluates whether systems can anticipate human movement well enough to share a space with people.
Education
03- 2024–Present
PhD, Computer Science
Georgia State University · Atlanta, Georgia
AMAI Research Lab, advised by Dr. Haoxin Wang. Expected May 2028.
- 2020–2021
MS, Business Analytics
University of Arizona, Eller College of Management · Tucson, Arizona
- 2016–2020
BS, Mathematics
University of Arizona · Tucson, Arizona
Positions
04- 2024–Present
PhD Student, AMAI Research Lab
Georgia State University · Atlanta, Georgia
- Advised by Dr. Haoxin Wang.
- Designing PoseBench, a benchmark that isolates interaction reasoning from single-person motion modelling in multi-agent pose prediction; in preparation.
- Developing a frequency-domain method that separates interaction effects from individual motion in multi-agent pose prediction for human–robot teaming; under review, 2026.
- Co-authored EpiFormer (arXiv, June 2026), an early-fusion geometric deep learning model for antibody epitope prediction that lifts F1 on the AsEP benchmark from 0.337 to 0.482.
- Co-authored LM-Meter (ACM/IEEE SEC, December 2025), a lightweight on-device profiler that exposes phase- and kernel-level inference latency of language models at under 3% throughput cost.
- 2022–2023
FSO Technology Consultant, Emerging Technology
EY · Phoenix, Arizona
- Designed and built an AI proof of concept that generates PowerPoint decks automatically, streamlining content creation and lifting team productivity.
- Kept the prototype current with advances in AI and emerging technology, and proposed the next round of capabilities.
- 2021
Student Consultant Leader
University of Arizona, Eller College of Management · Tucson, Arizona
- Led a student consulting team for a leading mining company through data collection, preparation, and analysis, delivering five analyses and two new KPIs.
- Built a Power BI drill-down KPI for haul-cycle productivity and a detractor model that traces the causes of low productivity.
- Recommended KPIs to raise operating efficiency and lower operating costs.
- 2019
Undergraduate Research Assistant
Yao System Biology Lab, University of Arizona · Tucson, Arizona
- Analysed and visualised cancer-cell life-cycle data in Python, R, and Tableau, surfacing trends that lab researchers built on in their own work.
- Collected and cleaned research data across several lab projects.
Awards & honours
03Academic Distinction of the Academic Year
University of Arizona
Dean's List with Distinction
University of Arizona
Dean's List
University of Arizona
Teaching & mentoring
01- 2021
Graduate Teaching Assistant
University of Arizona, Eller College of Management · Tucson, Arizona
Service
02- Present
President, ACCC Faith Hope Love Fellowship (GSU Branch)
Georgia State University · Atlanta, Georgia
- 2019–2020
Treasurer, RiskRunners Actuarial Science Club
University of Arizona · Tucson, Arizona
- Ran weekly club activities, study groups, online Q&A sessions, and networking events that connected students with professionals.
- Represented the University of Arizona at the WSIA Actuarial Symposium at Arizona State University, fall 2019.
Recent publications
View all →EpiFormer: Learning Antigen-Antibody Interactions for Epitope Prediction via Geometric Deep Learning
arXiv preprint arXiv:2606.04154 · 2026
LM-Meter: Unveiling Runtime Inference Latency for On-Device Language Models
The Tenth ACM/IEEE Symposium on Edge Computing (SEC ’25) · 2025