Design and architect large-scale GPU clusters across compute, networking, and storage. Model performance for AI/ML workloads, define low-latency interconnects (InfiniBand/RoCE), integrate storage, and operationalize reliability and monitoring with SRE, networking, and storage teams.