AI Infra Architect

Dr. Leo is an Agent Army Architect and AI infrastructure innovator working at the intersection of AI systems, networking, and high-performance computing. His work explores Guided AI Engineering—a practical approach combining human-in-the-loop development with structured workflows to make AI systems more reliable for real-world engineering.

Recently, his focus has expanded toward agent-native applications and infrastructure ecosystems, with an emphasis on Agent Runner and the mailbox protocol—a shift from request/response APIs to persistent, message-driven agent coordination. In this model, software acts as a routing and orchestration layer, enabling scalable, asynchronous collaboration across large numbers of agents.

Dr. Leo has 15+ years of experience in HPC and distributed systems, with expertise in GPU clusters, CUDA, MPI, RDMA, FPGA, and PCIe fabrics. He earned his PhD from the University of Wisconsin–Madison, has worked with Cisco and multiple startups as founder and CTO, and holds 21 patents in networking and HPC architecture.

Currently, he works as an AI Infrastructure Architect in Oracle’s Center of Excellence, focusing on benchmarking and architecting large-scale AI infrastructure across RDMA fabrics, NVLink systems, and OCI superclusters.