Worked on the ecmwf/anemoi-utils repository to improve dependency management and error handling in Python. Focused on making MLflow an optional dependency by introducing conditional imports, ensuring that the core functionality remains unaffected when MLflow is not installed. Enhanced user experience by implementing explicit error messages that clearly inform users about missing dependencies. Updated the packaging to include the requests library as part of the MLflow optional dependencies, supporting new MLflow-related submodules. This approach reduced potential runtime errors and improved maintainability, while preserving backward compatibility and minimizing the impact on the existing codebase during the month’s development cycle.