This document presents a proposed peer-assisted video sharing system for online social networks. It begins with an introduction to increased popularity of video sharing and existing peer-assisted video on demand techniques. It then describes limitations of current client-server architectures, including high bandwidth costs. The proposed system aims to improve performance by exploring social relationships and interest similarities between peers. It uses four algorithms to construct peer overlays, prefetch chunks, deliver chunks and scheduling. The implementation aims to reduce data storage and bandwidth requirements while providing scalability. Simulation results show improved startup delays and reduced server traffic compared to existing systems.