Create inspectable infrastructure evidence when production systems cannot be shared.

Projects should demonstrate provisioning, failure handling, observability, automation and capacity reasoning.

Kubernetes cluster

For kubernetes cluster, focus on where it sits in the system, what it depends on, how failure becomes visible, and what evidence would show you can reason about it in the context of AI Infrastructure Portfolio & Homelab Projects.

GPU scheduling simulation

For gpu scheduling simulation, focus on where it sits in the system, what it depends on, how failure becomes visible, and what evidence would show you can reason about it in the context of AI Infrastructure Portfolio & Homelab Projects.

Bare-metal provisioning

For bare-metal provisioning, focus on where it sits in the system, what it depends on, how failure becomes visible, and what evidence would show you can reason about it in the context of AI Infrastructure Portfolio & Homelab Projects.

Network lab

Distributed AI is dominated by east-west traffic between accelerators and nodes. Learn leaf-spine architecture, routing, MTU, congestion, loss and telemetry before going deeper into RDMA, RoCE or InfiniBand. NVIDIA current reference architectures use dedicated high-bandwidth fabrics because network behavior directly affects distributed workload efficiency.

Observability stack

Useful observability spans workload, scheduler, GPU, host, network and storage. The skill is correlation: deciding whether a slow training job comes from compute throttling, a failed link, queue pressure, storage saturation or an application-level issue.

Capacity project

For capacity project, focus on where it sits in the system, what it depends on, how failure becomes visible, and what evidence would show you can reason about it in the context of AI Infrastructure Portfolio & Homelab Projects.

Incident postmortem

Use a structured incident process: detect impact, preserve evidence, stabilize service, isolate the failure, restore carefully, then document contributing factors and corrective actions. A restart without diagnosis is not a postmortem.

The strongest preparation for AI infrastructure projects is a combination of system understanding and inspectable evidence: a design note, lab, automation workflow, benchmark, incident analysis or capacity model that you can explain under questioning.

Sources