During March 2026, contributed to the PaddlePaddle/GraphNet repository by developing a new computational graph for the PIGU_Hybrid model, expanding framework capabilities and supporting more flexible experimentation. This work involved Python and PyTorch for model development and data processing, as well as adjustments to ensure continuous integration compatibility and reliable builds. Additionally, improved the contributor experience in the PaddlePaddle/docs repository by correcting outdated API guide links, enhancing documentation reliability and navigation. The combined efforts reduced build friction, stabilized CI processes, and streamlined onboarding for external contributors, reflecting a focus on both technical depth and open-source collaboration best practices.