ML Products have become a prolific and integral part of taking the insights of Data Science from theory to reality. Oddly though, the path from conception to implementation is often unclear with seemingly few similar examples to work from. The result is often a sea of agony between sliding deadlines, heroic efforts of people working though unforeseen challenges and haphazard innovation. Each time a beautiful model makes its impact on the business bottom line, something worked. In this talk we present the ML Playbook. It pulls together the best aspects from a variety of successful ML Product launches into a cohesive strategy to Plan, Build, Test, Learn, and Release ML Products. We'll demonstrate the ML Playbook in action with the story of launching an alert monitoring product for the world's most powerful jet engines, the GE90-115B.