Worked on enhancing video metadata extraction for the yt-dlp repository, focusing on Bluesky video content. Developed improvements that enable the extractor to prefer the source video format, reducing ambiguity and increasing data reliability. Addressed metadata accuracy by fixing MD5 hashes and uploader information for specific Bluesky posts. Expanded the metadata schema to include chapters and heatmap fields, supporting richer downstream analytics and improved search relevance. Utilized Python for API integration, data extraction, and web scraping, demonstrating a methodical approach to refining data pipelines. The work delivered a targeted feature with depth, emphasizing data quality and extensibility within a short timeframe.