During a two-month period, Xpf6677 contributed targeted backend improvements to intelligent-machine-learning/dlrover and HKUDS/LightRAG. In dlrover, Xpf6677 enhanced developer productivity and integration quality by correcting a class name typo in RayJobSubmitter and updating build scripts to generate Protocol Buffers .pyi stubs, improving type hinting and IDE support. For LightRAG, Xpf6677 resolved an initialization issue by ensuring the initialize_rag function returns the Rag object after storage backends are set up, preventing downstream errors. These contributions, implemented in Python and Shell, demonstrated careful attention to code correctness, maintainability, and documentation, resulting in more stable and developer-friendly codebases.