I suggest it because that's exactly how I put my first ML model into production a couple of weeks ago. All my model does is update placeholder text to give suggestions to users based on previous actions. Took about 3 hours from start to finish: write training script, run it on cron, throw result in a cache, add API endpoint for accessing cached model, fetch model on frontend boot and store in state store, and integrate with search component.

https://updates.moonlightwork.com/updated-skill-search-now-w...

This book is kind of dated, but it's what I used for inspiration and reference: https://www.amazon.com/gp/product/B00F8QDZWG/ref=oh_aui_d_de...