04·ML project lifecycle: data → training → evaluation → deployment

Introduction

Contrary to appearances, model training is only about 20% of the work on an ML project. The rest is: collecting and cleaning data, checking that the model does not cheat, deploying to production, monitoring whether it degrades over time. In this lesson you will learn the entire ML project lifecycle — from idea to a working system in users’ hands. This is the knowledge missing from most "AI courses" that focus only on training itself, and it decides whether a project actually delivers value or ends up on a shelf.