Practical experience working with observability systems (e.g., Prometheus, Grafana, Loki, or similar), including building custom exporters or collectors, setting up alerts, and handling metric cardinality and retention on a large scale. The role has a complete, end-to-end understanding of the HPC/AI system, running and interpreting microbenchmarks and workloads to confirm system readiness, then establishing the observability that maintains this state.