Creating Inference Services using xpresso.ai

Once a Machine Learning model has been trained, it needs to be deployed, where it can be used to make predictions based on new data.

The best practice for this deployment is to create a REST API for the model, referred to as an Inference Service. Clients requiring predictions on data then make calls to this REST API. Optionally, a GUI can also be created for this purpose.

xpresso.ai provides support for creating Inference Services through a special type of solution component called inference_service.

Developers need to create an inference_service component for each of the models they wish to deploy. The skeleton code created for this type of component during the build process has the infrastructure for the REST API already included. The developer has to implement minimal code in order to incorporate the model prediction logic into the service.

This development process is documented in this section.