PhD student @ UW

I am a researcher developing robust and accessible AI systems for healthcare. My work spans the theoretical foundations of machine learning robustness as well as the design, development, and deployment of practical AI tools that support healthcare delivery and education.

I study the robustness of machine learning systems under distribution shifts and develop libraries for improving their reliability in real-world deployment. I also build and deploy resource-efficient clinical decision-support tools to edge consumers for healthcare delivery and education, with an emphasis on efficiency, accessibility, and on-device deployment. My broader goal is to translate advances in machine intelligence into accessible and scalable systems that augment people’s wellbeing.

Before joining UW, I worked on reducing medication burden among older adults with multimorbidity by developing a case-control retrieval system upon electronic health records, advised by Prof. Tjeerd van Staa. I also studied disease subtyping in intensive care settings by learning representations of multimodal longitudinal patient trajectories, advised by Dr. Xiaxia Yu.

In my spare time, I enjoy hitting the weights, tinkering with electronics, and exploring new things on GitHub.

Current Projects

• DeconDTN: Developing algorithms and open-source tools to stress test and improve the robustness of machine learning systems trained on heterogeneous data sources before production.
• COACH: Building digital navigators that provide accessible coaching support for digital mental health interventions through distilled language models deployed on edge devices.
• Reflect.AI: Developing on-device clinical support tools that enhance patient–clinician communication during telehealth visits through real-time speech recognition and language-model-assisted feedback.