Worked on the kvcache-ai/Mooncake repository to deliver performance and reliability enhancements for SPDK NoF (NVMe over Fabrics) storage operations. Developed a dedicated worker pool to increase throughput and scalability, while implementing routing for NoF replicas to maintain contiguous data slices and improve data integrity. Introduced a benchmarking suite to enable data-driven optimization of the worker pool. Updated memory allocation strategies to support SPDK DMA capabilities, ensuring efficient data transfers. Enhanced error handling and reporting for replica transfers, improving system observability and resilience. The work focused on C++ development, leveraging expertise in benchmarking, memory management, and data transfer protocols.