Aggregate, process, clean, label, and version data, write and debug model code, provision compute (somehow find compute), run experiments, review results, deploy model, monitor predictions and close data flywheel loop.

머신러닝 애플리케이션 개발의 이상과 현실

Untitled