Worked on the temporalio/temporal repository to enhance observability by developing a per-queue tasks_dropped metric for the matching service, enabling detailed tracking of backlog task drops with reason tagging such as internal_error, data_loss, and expired_read. Leveraged Go for backend development, focusing on metrics instrumentation and robust unit testing to ensure code quality and operational reliability. Addressed Prometheus label conflicts by reverting the initial rollout to maintain nightly build stability, while coordinating with teams to plan a consistent label schema for future reintroduction. The work improved operational visibility, supporting faster incident diagnosis and more effective backlog management across distributed task queues.