Worked on the nebius/soperator repository to enhance reliability and safety in Kubernetes operations by delivering two features and resolving one bug. Focused on improving backup configuration through Helm, enabling clearer and more maintainable backup processes by allowing control over skipping backups without annotations. Introduced drift detection across all HelmRelease resources with configurable notification modes, adding default safety measures to prevent deployment drift. Addressed cluster configuration by ensuring workerFeatures in SlurmCluster defaulted to sensible values, reducing misconfiguration risk. Leveraged DevOps practices and expertise in Kubernetes and YAML to align technical solutions with operational goals for safer, observable, and scalable deployments.