In part 3 of this series, I dig into some detail about how I further parallelized the work and reduced overall runtime.