Worked on stabilizing the test infrastructure for Airflow 3 within the aws/amazon-mwaa-docker-images repository, focusing on improving CI reliability and test script execution. Addressed a compatibility issue by implementing a conditional approach to resolving remote task log group ARNs and the execution API server URL, ensuring these values are only processed if present in the configuration. This Python-based solution, leveraging configuration management and DevOps practices, prevented unnecessary environment variable handling during test setup. As a result, the changes reduced CI flakiness and setup time, supporting more reliable validation and smoother adoption of new Airflow 3 images by downstream users.