For years, forming student teams was a manual process driven by intuition and prone to failure. This project aimed to operationalize that intuition, replacing guesswork with an optimization engine designed to minimize social friction and maximize peer learning. The project spans two distinct implementations. It begins with a prototype that validated a core psychological hypothesis: that **skill homogeneity** mitigates social friction in project-based learning environments. It concludes with a deep dive into systems engineering, detailing how I rewrote the core engine using a modern systems programming language to tackle critical performance bottlenecks. The following report documents the complete project lifecycle, from the initial behavioral analysis and mathematical formalization to the final systems-level implementation. It illustrates how bridging the gap between high-level domain logic and low-level memory optimization creates a robust, scalable solution to a complex resource allocation problem.