There's three main components to Lukai you'd be interacting with.
- Web Interface
- Client Library
- Python Model Creation Library
The web interface provides all the feedback and information you need about the trained model.
The client takes care of all training and model deployment. It's designed to be used as a black box that you can log your training data into and get predictions out. It also provides hooks so you can tell it when it's allowed to train, to avoid impacting users.
The client library is written in Go, but it's easy to use it in Java (Android) and Objective-C/Swift (iOS).
This library is used by the data scientists creating the models. It's a small library that packages up your model and launches a job to train it. It only takes a handful of lines of code to integrate and use.