Worked on the DataDog/integrations-core repository to enhance observability for Kafka partition metrics by introducing an out_of_sync_broker_id tag. This feature enables identification of replicas not in the in-sync replica set, allowing teams to attribute under-replicated partitions to specific broker IDs for more targeted troubleshooting. The implementation, developed in Python with a focus on backend development and monitoring, preserves compatibility with existing dashboards and alerts for fully replicated partitions. Test coverage and code style were improved in response to feedback, ensuring reliability and maintainability. The work demonstrates depth in Kafka monitoring and Python testing within a production observability context.