Worked on enhancing the reliability of JDBC sink pipelines in both the apache/pulsar and datastax/pulsar repositories by implementing robust fatal error handling and state management. Developed a system in Java that ensures proper termination of sink instances when unrecoverable errors occur during flush operations, reducing the risk of stale or dangling processes. Ported these changes across forks to maintain consistent behavior and operational safety. Focused on backend development and unit testing to validate the new lifecycle management, this work reinforced data integrity and cross-repository code reuse, contributing to a safer and more predictable sink lifecycle for Pulsar-based data pipelines.