About Me
I am a PhD candidate in Computer Science at Rice University, working broadly in machine learning and artificial intelligence, with particular interests in the mathematical foundations of deep learning, generative modeling, and AI-driven scientific discovery. My research spans theoretical analysis, model and algorithm development, and large-scale experimentation, with much of my recent work focusing on applications in the physical sciences.
My current research focuses on architectural modifications to diffusion and flow matching focused on novelty of generations and computational efficiency. More broadly, my research group is exploring LLM-guided scientific discovery, with an emphasis on integrating generative models, predictive methods, and scientific reasoning. Previsously, I worked on Transformer variants and MLIPs. A recurring theme of my work is understanding how mathematical and physical constraints can be incorporated into learning systems without limiting their expressivity.
My work is highly interdisciplinary, involving close collaboration with researchers across machine learning and the physical sciences. I particularly enjoy bridging these fields, both by translating scientific problems into machine learning formulations and by communicating technical ML concepts to scientific audiences. I am advised by Dr. Anastasios Kyrillidis and Dr. Geoffroy Hautier, and collaborate with Dr. Christopher Jermaine and Dr. Thomas Reps.
Tutorials and Lecture Notes
Diffusion and Related
Transformers and Related
- Lecture Slides on Introduction to Transformers (pdf)
- Reinforcement Learning from Human Feedback Introduction (pdf)
- Transformer Mathematics In Extensive Detail (pdf)
- Deep Dive Into Attention Computations (pdf)