The most latency sensitive parts are written in C++. Due to our interconnected services architecture, we use either Python or C++ for each service, with the performance critical parts being C++14. C++