I built this to explore low-latency C++ patterns.

It uses a sharded architecture (one thread per symbol group) to avoid mutex contention during matching. Recently moved to int64_t fixed-point pricing and added memory compaction to prevent leaks in long-running processes.