Worked on the roboflow-python repository to enhance the stability and reliability of annotation loading, focusing on bug fixes rather than new features. Addressed issues with YAML dictionary-style label maps, ensuring correct parsing even when integer keys are present, and introduced targeted tests to validate these improvements. Leveraged Python and YAML for robust data loading and file parsing, while maintaining strong version control practices. Released a maintenance patch, updating the package to version 1.2.1 to reflect these changes and maintain downstream compatibility. The work emphasized thorough testing and careful handling of edge cases to improve the robustness of dataset annotation workflows.