Hyunjin Seo

Hi! I'm a Ph.D. student at KAIST, advised by Prof. Sungsoo Ahn.
I'm broadly interested in applying geometric deep learning to proteins and drug-like molecules.

Previously, I worked on GNNs in more general graph domains, combining them with LLMs for graph reasoning, and exploring pruning and calibration techniques to improve their efficiency and reliability.

More recently, I’ve been drawn to the AI4Science domain. I developed an energy-based graph transformer to predict ground-state molecular conformations, and later built a masked diffusion model for molecules with element-wise diffusion trajectories. Currently, I'm collaborating with Prof. Gyurie Lee on dynamics-aware protein design, modeling conformational changes in proteins induced by ligand binding.