Research

My research interests include trustworthy machine learning, ML fairness, robust evaluation of AI systems, human-AI interaction, and the social effects of increasingly personalized and socially interactive AI.

I'm especially interested in how AI systems behave in high-stakes or emotionally significant contexts, and how technical evaluation can better account for real users and social consequences.

Featured Research

Mitigating Dialect Bias in Toxicity Detection

Investigated disproportionate false-positive toxicity classifications for African American English and evaluated fairness interventions — including vector scaling, reweighting, and adversarial debiasing — reducing the AAE–SAE false-positive-rate gap by 95%.

Responsible AI ML Fairness NLP Model Evaluation
Read full case study →

Human-AI Interaction

LLM-Assisted Debugging for Novice Programmers

Studied how novice programmers interpreted, trusted, and acted on AI-generated debugging explanations from a Python-based LLM assistant with a REPL-style interaction flow.

Human-AI Interaction Trust in AI Usability Evaluation LLMs
Read case study →

Prior Human-Centered Research

Happy to Be Included

Designed and developed an inclusive job-recruitment platform, translating stakeholder research and usability findings into product requirements. Brave Award, UW HCDE Capstone Showcase.

Read case study →

Research in Progress

In Progress

I am currently contributing to early-stage research on persona inference in conversational language models and the performance and safety of intimacy in human- and AI-operated conversations.

This work is ongoing. Methods, results, and publication status are not yet finalized, so it does not yet have a dedicated project page.