GPU Infrastructure Architect – GPU Cluster Stand-up, Configuration & Operations (AMD MI350)

The GPU Cluster Architect is responsible for designing, provisioning, and operating AMD MI350–based GPU clusters on a cloud platform. The role ensures scalable, secure, and reproducible GPU infrastructure to support distributed training and high-performance workloads.

 

Key Responsibilities

• Design end-to-end GPU cluster architecture covering compute, networking, storage, and control services.
• Provision and operationalize up to 9 AMD MI350 GPU clusters based on confirmed cloud SKU availability.
• Configure GPU compute nodes including base OS images, GPU drivers, runtime libraries, and distributed training dependencies.
• Implement automation for node imaging, bootstrapping, lifecycle management, patching, and upgrades.
• Standardize environments using reproducible builds and Infrastructure-as-Code (IaC).
• Enable workload portability through containerized environments and documented deployment patterns.
• Implement OS baseline hardening, restricted administrative access, and secure cluster access controls.
• Establish monitoring, logging, and operational runbooks to ensure reliability and performance.