Worked on the kvcache-ai/Mooncake repository to deliver static labeling for client metrics, enabling tagging by host IP and cluster name for improved per-host and per-cluster observability. Implemented cluster identification using MC_STORE_CLUSTER_ID to support LLM clusters and performed targeted performance optimizations in the metric reporting path by removing unnecessary move operations. Addressed test case errors to stabilize metric reporting, which enhanced dashboard reliability and reduced mean time to resolution for metric-related issues. Leveraged C++ development skills with a focus on metric reporting and performance optimization, resulting in clearer monitoring, more accurate resource attribution, and increased productivity for developers and operations teams.