Continuous integration and continuous deployment

In machine learning, we tend to look at two somewhat different stacks. On the one hand, you have the model creation and deployment process. This includes your model artifacts, datasets, metrics, and target deployment options. As we discussed previously, you might create a pipeline to automate this. On the other hand, you have the actual software application where you want to expose your model. This might be a visual search mobile app, a question/answering chat, an image generation service, a price forecasting dashboard, or really any other process to improve using data and automated decisions.

Many software stacks use their own continuous integration and continuous deployment (CI/CD) pipelines to seamlessly connect all the parts of an application. This can include integration tests, unit tests, security scans, and machine learning tests. Integration refers to putting the application together, while deployment refers to taking steps...