I am a Research Scientist at IBM Research, working with the MIT–IBM Computing Research Lab. I am also an ML PhD student at Georgia Tech, where I was awarded the NSF GRFP and am advised by Polo Chau.
I work on generative models, combining theory, interpretability, and software engineering to better understand them and make them more capable and controllable. I have a particular interest in diffusion and flow-based generative models, including their theoretical foundations (e.g., What Time Is It?), internal representations (e.g., ConceptAttention), and applications to language and reasoning (e.g., Flow Reasoning Models).
I also use data visualization to make complex models easier to understand and explore. I build interactive tools, open-source software, and animations for researchers and broader audiences (e.g., Diffusion Explorer, ManimML).
I have worked across academic and industry research at IBM Research, the MIT–IBM Computing Research Lab, Apple, Adobe Firefly, NASA JPL, and the University of Pittsburgh, spanning language, multimodal safety, scientific visualization, and computational biology.
-
Started as a Research Scientist at IBM Research, working with the MIT–IBM Computing Research Lab.
-
What Time Is It? was accepted to NeurIPS 2026.
-
Flow Reasoning Models was accepted to the DiffuLM workshop at NeurIPS 2026.
-
The Rectified Flow Explainer was accepted to the VISxAI workshop at IEEE VIS 2026.
-
Started as a research intern at the MIT–IBM Computing Research Lab.
-
Passed the Georgia Tech Machine Learning PhD qualifier; my visual write-up covers flow matching and rectified flow.
-
SafetyPairs was accepted to the Trustworthy AI workshop at ICLR 2026.
-
Transformer Explainer was accepted to CHI 2026.
-
LORE was accepted to ICLR 2026.
-
Diffusion Explorer was accepted to IEEE VIS 2025.
-
Started as an AI research intern on Apple's Responsible AI team.
-
ConceptAttention was selected for an oral presentation at ICML 2025.
-
ConceptAttention won Best Paper at the CVPR Workshop on Visual Concepts.
-
Transformer Explainer was presented at IEEE VIS 2024 and won Best Poster.
-
ClickDiffusion was accepted to the AI for Content Creation workshop at CVPR 2024.
-
Received the National Science Foundation Graduate Research Fellowship.
-
LLM Self Defense was accepted to the ICLR 2024 Tiny Paper Track.
-
Manifold Contrastive Learning was accepted to Transactions on Machine Learning Research (TMLR).
-
ObjectComposer was accepted to the NeurIPS 2023 Workshop on Machine Learning for Creativity and Design.
-
ManimML was presented at IEEE VIS 2023 and won Best Poster.
-
Started my Machine Learning PhD at Georgia Tech with a President's Fellowship.
Blogs
Visualizations
I am most interested in diffusion and flow-based generative models, including 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.