I am an ML PhD student at Georgia Tech advised by Polo Chau and supported by the NSF GRFP.
I work on diffusion and flow-based generative models — both their theoretical foundations (e.g., how data geometry lets flow matching models drop time conditioning) and their architectural and mechanistic properties (e.g., ConceptAttention, which shows diffusion transformers learn highly interpretable features). Most of my work to date has been on models for images, video, and multimodal understanding, and I am increasingly interested in their applications to language and reasoning. I enjoy combining theoretical insight with architectural design, and I also develop open-source software libraries and interactive tools and animations that make these models legible to a wider audience.
I have had the opportunity to work with many great people on a variety of topics in both academic and industry research labs: at IBM Research on natural language summarization, this coming summer at the MIT-IBM Watson AI Lab, at Apple on the Responsible AI team on improving vision-language models, at Adobe Firefly on improving the safety of text-to-image models, at NASA JPL on interactive visual tools for space science, and at the University of Pittsburgh on computational biology.
I am most interested in diffusion and flow-based generative models — their theory, their architectures, and their applications across vision, multimodal understanding, and language and reasoning. I also enjoy building open-source libraries and tools that make these models more accessible.