During a two-month period, Mingfei Gao contributed to both pinterest/ray and bytedance/deer-flow, focusing on backend and frontend improvements using Python, React, and TypeScript. For pinterest/ray, Mingfei optimized the Autoscaler Monitor by restructuring resource demand parsing, reducing redundant computations and improving autoscaling efficiency without altering existing behavior. In bytedance/deer-flow, Mingfei implemented automatic configuration reloading to streamline developer workflows and fixed a frontend rendering issue by refining SubtaskCard logic, enhancing UI stability. These contributions demonstrated a thoughtful approach to performance optimization, resource management, and testing, with changes that were well-contained, review-friendly, and impactful for development and operational reliability.