During October 2024, Andre Ramos developed a configurable local directory feature for model snapshot downloads in the google-research/timesfm repository. Leveraging Python and his expertise in data management and machine learning, he enabled users to specify custom storage locations for model artifacts, enhancing control over data locality and deployment workflows. Andre integrated the new configuration seamlessly into the existing snapshot process, ensuring backward compatibility and robust path validation. By linking the delivery to a specific commit, he maintained clear traceability for audit and review. This work improved reproducibility and flexibility for users managing machine learning models across diverse environments.