Developed and integrated a Most Recently Used (MRU) cache policy for the v1 storage backend in the LMCache/LMCache repository, focusing on improving cache predictability and backend utilization. The work involved designing the MRU policy within the existing caching framework, wiring it into the cache policy factory, and ensuring compatibility with pinned items. Comprehensive tests were added to validate correct eviction behavior and lifecycle management of cache entries. Using Python and leveraging skills in cache management, software design, and testing, the implementation enabled more predictable eviction of hot data and reduced cache churn, supporting future policy experiments and performance tuning.