Contributed to the apache/flink-cdc repository by enhancing data handling and serialization reliability for streaming data pipelines. Delivered a feature in Java that enables coercion of complex types such as Array, Map, and Record to STRING, supported by comprehensive unit and integration testing. Addressed serialization issues in Debezium JSON for Kafka sinks, ensuring correct mapping of non-string default values to Java types. Improved codebase maintainability through refactoring and alignment with Java packaging standards. Further strengthened schema evolution robustness by refining decimal type merging logic to respect precision constraints, reducing runtime errors and improving the reliability of change data capture workflows.