My research develops mathematically grounded methods to improve the efficiency and robustness of Optimal Transport, which are crucial for trustworthy and adaptive models in real-world AI applications. I am also particularly interested in extending these techniques to non-Euclidean settings, which opens up possibilities for cross-domain translation and representation learning with more accurate geometric characterizations. I see strong potential for synergy between my theoretical work and VALIANT’s emphasis on immersive and translational AI, especially in developing foundational tools that can be applied across disciplines such as healthcare, neuroscience, and beyond.