I'm motivated by the structure underneath learning — the optimization landscapes, the geometry of graphs, and the architectures that might one day think a little more like a brain.
Every learning algorithm is, at heart, an optimization problem. I'm drawn to how problems are framed — linear and convex formulations, matrix structure, and the trade-offs between objectives like L1 and L2 — and to what those choices imply for robustness and generalization.
Graphs encode relationships the way the world actually works. I study how information propagates across them and where message passing breaks down, looking for architectures that respect graph structure rather than fighting it.
I'm optimistic that neuroscience-inspired models are indispensable to genuinely human-like AI. Bridging biological principles of computation with deep learning is, to me, one of the most exciting frontiers in the field.
The mathematics beneath the models. Spectral properties, connectivity, and combinatorial structure give graph learning its theoretical backbone — and I like working from that foundation upward.
I'm early in my research journey and actively building toward my first contributions. When papers and preprints land, they'll appear here in full.