Architecting High-Performance AI/ML Infrastructure at the Intersection of GPU Compute, RDMA, and Enterprise Storage.