Worked on the agno-agi/agno repository to deliver a document processing enhancement focused on expanding file type support and improving metadata extraction. Developed a feature that added JSON and DOCX MIME type recognition to the file validation logic, enabling the system to handle a broader range of documents. Refactored the agent run creation process to ensure accurate propagation of file metadata, including filename, content, and MIME type, for injected files. Leveraged Python for backend and API development, with an emphasis on robust file handling. The work improved downstream indexing and analytics by making richer metadata available throughout the data pipeline.