Worked on the ray-project/kuberay repository to enhance observability by adding the Custom Resource UID as a label to metrics for both RayCluster and RayJob resources. This update, implemented using Go and the controller-runtime framework, enables unique identification of each resource instance within Kubernetes monitoring systems. By improving the precision of metrics instrumentation, the work allows site reliability engineers to correlate metrics directly with specific custom resources, streamlining troubleshooting and monitoring workflows. The changes focused on metrics and observability, laying the foundation for more granular cross-resource monitoring and supporting operational visibility across Ray deployments without introducing new bugs.