Yevgeny Shnaidman enhanced GPU observability in the NVIDIA/gpu-operator repository by developing a PrometheusRule that translates DCGM metrics into user-friendly names and appends a vendor label for NVIDIA. Using Kubernetes and Prometheus, he focused on improving the clarity and discoverability of GPU telemetry within accelerator dashboards. His work addressed the need for actionable metrics by standardizing naming conventions and enriching metric context, which supports faster issue diagnosis and more effective capacity planning. The solution, implemented in YAML, laid a solid foundation for future service level indicators and objectives, demonstrating a thoughtful approach to observability and monitoring in cloud-native environments.