Understanding challenges with ML systems and projects
Implementing a product leveraging ML can be a laborious task as some new concepts need to be introduced in the book around best practices of ML systems architecture.
So far in this book, we have shown how MLflow can enable the everyday model developer to have a platform to manage the ML life cycle from iteration on model development up to storing their models on the model registry.
In summary, at this stage, we have managed to create a platform for the model developer to craft their models and publish the models in a central repository. This is the ideal stage to start unlocking potential in the business value of the models created. In an ML system, to make the leap from model development to a model in production, a change of mindset and approach is needed. After unlocking the value and crafting models, the exploitation phase begins, which is where having an ML systems architecture can set the tone of the deployments and operations...
