Deploying a model with managed online endpoints through the Python SDK v2

In the previous section, we deployed an MLflow model, but when you create a model that does not leverage MLflow, you need to provide two additional details for a successful managed online endpoint deployment. In this section, we will focus on adding functionality so that we can deploy our model without relying on MLflow to provide the environment and scoring script.

In order to deploy a managed online endpoint leveraging the SDK v2 and not relying on MLflow to provide the environment and scoring script, we will create those in this section as you leverage the notebook: Chapter 6 Model Deployment SDK V2.ipynb:

1. Our first step is to create our score.py file. This is the file that will be used to load the model and serve a request to the endpoint.

The following code snippet provides the information required for the entry script:

Figure 6.26 – The score.py script...