AWS Marketplace
User Guide for AWS Marketplace Providers

The AWS Documentation website is getting a new look!
Try it now and let us know what you think. Switch to the new look >>

You can return to the original look by selecting English in the language selector above.

Machine Learning Products

AWS Marketplace enables sellers to create and provide machine learning algorithms and model packages using Amazon SageMaker. Sellers package their products as Docker containers, upload them to Amazon Elastic Container Registry (Amazon ECR), create the algorithm or model packages in Amazon SageMaker, and add them as free or paid products in AWS Marketplace.

AWS customers can find these products through the Amazon SageMaker console or AWS Marketplace, and deploy them on Amazon SageMaker. They can review product descriptions, documentation, customer reviews, pricing, and support information. When the buyers subscribe to an algorithm or model package, the product is added to their list of products on the Amazon SageMaker console. They can also use the Amazon SageMaker SDK, the AWS Command Line Interface (AWS CLI), or the Amazon SageMaker console to create a fully managed inference endpoint. Buyers can access models only through the RESTful endpoints.

For support with creating machine learning products with Amazon SageMaker, contact AWS Marketplace Seller Operations.

Getting Started with Amazon SageMaker

If you are new to Amazon SageMaker, the following webinars can get you started:

Amazon SageMaker Algorithms and Model Packages

As a seller of Amazon SageMaker products, you can list an algorithm, a model package, or both. 

Amazon SageMaker Algorithm

An Amazon SageMaker algorithm enables buyers to perform end-to-end machine learning. It has two logical components: training and inference. Buyers use the training component to create a training job in Amazon SageMaker and build a machine learning model. Amazon SageMaker saves the model artifacts generated by the algorithm during training to the buyer's Amazon Simple Storage Service (Amazon S3) bucket.

Buyers use the algorithm’s inference component together with the model artifacts to build a model package which is then used to run real-time or batch transform jobs in Amazon SageMaker. As a seller, you can charge buyers for training and inference separately. 

Amazon SageMaker Model Package

Model packages contain a pre-trained model that buyers can use to run real-time or batch inference jobs in Amazon SageMaker. They use the model for hosting services or running batch transforms in Amazon SageMaker. A model package contains an inference component that is packaged along with model artifacts that you provide. As a seller, you can build your model artifacts by training with Amazon SageMaker or you can use your own model artifacts from a previously built model. You can charge buyers for the inference jobs.

For more information on how to put your algorithms and models on the AWS Marketplace, see Putting Your Algorithms and Model Packages on the AWS Marketplace.