Worked on reliability improvements for the EnterpriseDB/cloudnative-pg repository, focusing on the bootstrap process and its integration with the metrics-exporter. Addressed a recurring failure mode by restructuring the bootstrap logic to use two separate transactions, isolating the metrics-exporter setup from the primary replication bootstrap. This backend development effort, implemented in Go, enhanced fault isolation and ensured that errors in metrics-exporter configuration no longer impacted streaming replica creation or pod restarts. Demonstrated strong database management and transaction handling skills while collaborating across teams to deliver a production-ready fix that improved startup predictability and reduced downtime during cluster initialization and recovery.