Machine learning (ML) projects require a significant, multi-stage effort that includes modeling, implementation, and production to deliver business value and to solve real-world problems. Numerous alternatives and customization options are available at each step, and these make it increasingly challenging to prepare an ML model for production within the constraints of your resources and budget. Over the past few years at Amazon Web Services (AWS), our Data Science team has worked with different industry sectors on ML initiatives. We identified pain points shared by many AWS customers, which originate from both organizational problems and technical challenges, and we have developed an optimal approach for delivering production-ready ML solutions.

This guide is for data scientists and ML engineers who are involved in ML pipeline implementations. It describes our approach for delivering production-ready ML pipelines. The guide discusses how you can transition from running ML models interactively (during development) to deploying them as a part of a pipeline (during production) for your ML use case. For this purpose, we have also developed a set of example templates (see the ML Max project project), to accelerate the delivery of custom ML solutions to production, so you can get started quickly without having to make too many design choices

https://docs.aws.amazon.com/prescriptive-guidance/latest/ml-production-ready-pipelines/welcome.html