Memory management, particularly virtual memory, is a crucial topic in Operating System curricula. This paper explores the integration of CUDA Unified Memory into OS coursework to enhance student understanding of these concepts. By using GPU CUDA Unified Memory as a practical example, students were given a hands-on homework assignment containing several experiments designed to deepen their comprehension of memory management techniques. The experiments included observing page faults, techniques for reducing page faults, and oversubscription of GPU memory. This paper details the design rationales behind the hands-on experiments, aiming to provide a comprehensive learning experience. Pre-test and post-test assessments indicate that the assignment improves student learning outcomes, demonstrating the effectiveness of this pedagogical approach.