Worked on enhancing the GEPA optimization capabilities within the mlflow/mlflow repository by developing a feature that allows advanced keyword arguments to be passed directly to the GepaPromptOptimizer. This update improved the configurability and experimentation potential of GEPA optimization workflows, enabling users to access parameters previously unavailable through the MLflow interface. The implementation focused on Python development and machine learning, with careful attention to code quality and maintainability through a clean commit and sign-off process. Unit testing was used to reinforce stability, and the work aligned with project guidelines, supporting broader experimentation and parameter exploration for researchers using MLflow pipelines.