Who is Caikit for?

While there’s no prescribed roles for using or working with Caikit, the project is designed to satisfy two key audiences:

• Model authors
• Model users

Model user

Model users are, simply, the people who use models. They use inference-trained AI models to perform deeper analysis on data and to support AI-enabled applications. While there’s no reason why you couldn’t both create and consume a model, for Caikit, a model user is the consumer of the model results. You, as a model user, use caikt to turn data into actionable insights. For example, by using a sentiment analysis model to analyze information from a dataset of customer feedback. The insights from this analysis are your concern, while Caikit handles the lower-level infrastrucutre details that enable you to reliably inteact with the model.

Model users can range in roles from app developers to data scientists to service users.

Model author

Model authors are the people who build and train models for use in data analysis and AI-enabled applications. If you’re a machine learning engineer or a data scientist who’s interested in creating your own model (no matter what the complexity), this is you. Your are interested in creating, running, loading, deploying, and training models. To you, Caikit is an opportunity for consumers of your models to seamlessly integrate and maintain your model into their development stack.

Model authors bring data and tuning parameters to pre-existing algorithms and create new model artifacts.