Developer Guide

This guide provides an overview of the processes involved in developing and maintaining the Next Best Message, Search & Autocomplete feature. The detailed steps are divided into the following key areas, each described in its own sub-page:

1. Training Pipeline: This section explains the end-to-end process of preparing and training models, including data preprocessing, model training, evaluation, and deployment. For details, see the Training Pipeline page.

2. Training Data Selection: This section focuses on how training data is selected, cleaned, and prepared to ensure high-quality model performance. For details, see the Training Data Selection page.

3. Model Deployment and Serving: This section describes how trained models are deployed and served in production, including the automation and monitoring processes. For details, see the Model Deployment and Serving page.

4. Inference: This section explains how recommendations are created, detailing the sequence of operations and the components involved. For details, see the Inference page.

By following these structured processes, the system ensures that models are accurate, efficient, and continuously improved.