Developed a robust error handling and retry mechanism for the HKUDS/LightRAG repository, focusing on improving the reliability of large-batch ingestion and embedding workflows that interact with the OpenAI API. Leveraging Python for backend development, the work introduced retries for transient server errors and specific client-side issues, such as malformed request bodies, by defining a new exception type for transient bad requests. Enhanced exception handling was implemented to prevent error leakage, and comprehensive regression tests were added to validate the new logic. This feature reduced ingestion failures, resulting in more resilient API integration and streamlined processing for large-scale data operations.