ML Max is a set of example templates to accelerate the delivery of custom ML solutions to production so you can get started quickly without having to make too many design choices.
- ML Training Pipeline: This is the process to set up standard training pipelines for machine learning models enabling both immediate experimentation, as well as tracking and retraining models over time.
- ML Inference Pipeline: Deploys a model to be used by the business in production. Currently this is coupled quite closely to the ML training pipeline as there is a lot of overlap.
- Development environment: This module manages the provisioning of resources and manages networking and security, providing the environment for data scientists and engineers to develop solutions.
- Data Management and ETL: This module determines how the machine learning operations interacts with the data stores, both to ingest data for processing, managing feature stores, and for processing and use of output data. A common pattern is to take an extract, or mirror, of the data into S3 on a project basis.
- CICD Pipeline: This module provides the guidance to setting up a continuous integration (CI) and continuous deployment (CD) pipeline, and automate the delivery of the ML pipelines (e.g., training and inference pipelines) to production using multiple AWS accounts (i.e., devops account, staging account, and production account.).