Worked on stability and data integrity improvements for the columnflow/columnflow repository, focusing on backend development and data engineering using Python. Addressed two key bugs by refining axis existence validation in the CreateHistograms module, ensuring that None and empty axes were handled distinctly to prevent misclassification and improve reliability in edge cases. Additionally, corrected the stats output format from .parquet to .json, enabling proper JSON handling and downstream compatibility. These changes enhanced the reliability of histogram generation and data exports, demonstrating careful attention to edge-case validation, file I/O normalization, and clear commit documentation throughout the development process.