Worked on enhancing the AWS provider within the kubernetes/autoscaler repository by improving the accuracy of cluster autoscaling decisions. Focused on refining node group comparisons by ensuring the autoscaler ignores the topology.k8s.aws/zone-id label, which previously could cause erroneous scaling actions in AWS environments. Implemented this feature using Go, leveraging AWS and Kubernetes expertise to address edge cases where label values might be empty or differ between nodes. Added comprehensive test coverage to validate the new logic, ensuring robust handling of these scenarios. This work contributed to more reliable autoscaling behavior and aligned with business goals for AWS infrastructure management.