Worked on the kubernetes/autoscaler repository to enhance the CoreWeave autoscaler’s reliability and capacity planning capabilities. Addressed a bug in the node group deletion flow by refining the logic for node removal and correcting target size adjustments, ensuring accurate scaling and reducing the risk of under- or over-provisioning. Developed a new feature to simulate node provisioning using template node information, enabling more informed decisions about instance types for CoreWeave workloads. Leveraged Go, Kubernetes, and backend development skills to deliver improvements that support proactive capacity planning and faster incident response in production cloud environments, focusing on operational stability and accuracy.