Developed a policy-driven deletion strategy for RayJobs in the ray-project/kuberay repository, enabling resource cleanup actions based on job outcomes. This work involved updating Custom Resource Definitions (CRDs), enhancing API documentation, and extending controller logic to enforce new lifecycle semantics. Comprehensive tests were implemented to validate deletion policy behavior across both successful and failed job scenarios, ensuring reliability and reducing the risk of orphaned resources. The solution was built using Go and YAML, leveraging Kubernetes expertise and a focus on robust API design. These changes improved lifecycle management clarity for users and contributed to more predictable resource handling within the project.