Worked on enhancing data ingestion reliability for the opensearch-project/data-prepper repository by addressing a configuration issue in AWS Kinesis stream retrieval. Focused on correcting how retrieval configuration was applied to the Kinesis Client Library scheduler’s polling strategy, ensuring consistent and accurate propagation of settings throughout the scheduler lifecycle. Introduced a dedicated retrieval configuration holder to streamline scheduling behavior and facilitate future updates. Added targeted regression tests in Java to verify the fix and prevent similar issues. This work improved the stability and maintainability of Kinesis data ingestion, leveraging expertise in Java, AWS Kinesis, and general software development best practices.