Worked on stabilizing asset-triggered DagRun validation in the aws-mwaa/upstream-to-airflow repository, focusing on backend reliability and scheduler robustness. Addressed a DetachedInstanceError by ensuring both asset and alias loaders were applied before Pydantic validation, which reduced the risk of orphaned or retried tasks during asset-driven workflows. Extended the test suite to verify correct handling of source_aliases in callback payloads, improving coverage and reliability. Utilized Python, SQLAlchemy, and ORM techniques to align loader usage with validation paths, resulting in more predictable scheduler behavior. The work emphasized targeted bug fixing and thorough testing to enhance the stability of asset-driven DAG executions.