Worked on backend and frontend reliability improvements across microsoft/graphrag and langgenius/dify, focusing on critical bug fixes rather than feature development. In Graphrag, addressed a misconfiguration in the indexing workflow by updating the LLM endpoint configuration to load the api_base parameter, enabling flexible integration of alternative language models and reducing vendor lock-in. For langgenius/dify, resolved a URL parameter parsing issue that previously caused user identification errors during authentication redirects, ensuring accurate user_id extraction and stabilizing user flows. Utilized TypeScript and Python, applying skills in API integration and backend development to enhance system robustness and maintain code quality throughout.