Castilio worked on the Azure/telescope repository, focusing on enhancing cluster test observability by developing a feature for Cluster Loader 2 that enables Kubernetes metrics scraping through configurable flags. By integrating support for kube-state-metrics and metrics-server directly into the command construction, Castilio streamlined the process of collecting cluster state data during automated tests. This approach reduced manual setup and improved the speed and reliability of diagnosing cluster issues. The work demonstrated practical application of DevOps principles, Kubernetes monitoring, and Python scripting, resulting in more transparent and efficient validation of cluster states without introducing new bugs during the development period.