Develops theory-grounded, scalable algorithms for statistical learning and structured prediction. Focus on kernels, optimal transport and large-scale optimization, uniting provable guarantees with efficient approximations for modern, massive datasets.