Worked on the rungalileo/sdk-examples repository to enhance dataset experiments by standardizing the Message logging schema and normalizing enum casing, ensuring consistent and accurate logging of LLM messages. Addressed a critical bug by correcting Message and MessageRole imports and aligning enum values to system and user, which improved data integrity and reduced runtime errors. Leveraged Python for code refactoring, focused on enum usage, and updated import patterns to streamline maintainability. The work included thorough documentation and code reviews, resulting in improved data quality, reduced debugging time, and more reliable dataset experiment workflows without introducing new features during the development period.