From Model to Inference API

Day 51 of 100 Days of MLOps, and Module 6 begins. A trained model sitting in a file can't do anything — to serve predictions you wrap it in inference logic. Build that logic as a clean, HTTP-agnostic predict() function: validate the input, run the saved pipeline, return a structured result. Get this shape right and the web framework tomorrow is just a thin wrapper. Runs 100% locally on any OS.