Worked on refactoring Airflow-based data pipelines in the GoogleCloudPlatform/ml-auto-solutions repository to enhance code readability and maintainability. Focused on replacing ad-hoc task chaining with Airflow’s chain utility across multiple Directed Acyclic Graph (DAG) files, this approach clarified execution flow and simplified the process for code reviews and onboarding new engineers. Leveraged Python and data engineering best practices to standardize task sequencing, reduce pipeline complexity, and align with broader workflow improvement initiatives. The changes improved traceability and code quality while ensuring minimal risk, providing a clear rollback path and reinforcing reliability for the data engineering team’s ongoing development efforts.