Worked on the kubernetes-sigs/cloud-provider-azure repository to enhance node provisioning stability in Azure Kubernetes clusters. Developed a configurable grace period mechanism, using Go and Kubernetes APIs, to prevent premature node deletion when virtual machines are not yet visible in Azure Resource Manager. Introduced a toleration-based delay path for new nodes, reducing provisioning churn and improving autoscaling reliability. Migrated configuration from a hard-coded flag to a centralized Azure config property, increasing maintainability and operator flexibility. Expanded unit test coverage to validate the new grace period logic across multiple scenarios, ensuring robust handling of edge cases in cloud provider development workflows.