Worked on the DataDog/watermarkpodautoscaler repository to enhance Prometheus metrics labeling for the Watermark Pod Autoscaler, focusing on improving monitoring fidelity and autoscaling accuracy. The approach involved centralizing label generation and extending the default set of Prometheus labels, ensuring that all metric operations used complete and consistent label sets. This change addressed issues with incomplete label maps that previously led to silent no-op behavior in Prometheus, particularly in value deletion operations. The work was implemented in Go and leveraged DevOps practices within a Kubernetes environment, resulting in more reliable metric tracking and maintainable code for future enhancements and scaling decisions.