Developed and integrated a feature for the oumi-ai/oumi repository that enables saving evaluation configurations as YAML files within a specified output directory. This addition, implemented using Python and leveraging YAML for serialization, streamlined the evaluation workflow by reducing manual configuration steps and supporting reproducibility. The feature allows users to easily share and reuse evaluation setups, improving onboarding and collaboration for teams deploying evaluations. Work focused on backend development, ensuring seamless integration with existing processes without introducing new bugs. The approach emphasized maintainability and usability, enhancing the overall configurability of evaluation runs while aligning with established repository standards and practices.