Logging Models & Artifacts with Signatures

Day 35 of 100 Days of MLOps. A tracked run should carry more than numbers — the plots that explain it and a model that knows its own inputs. Log figures and files as artifacts, and save a model with a signature that records exactly what it expects, so it rejects bad inputs instead of silently guessing. This is what turns a run into a deployable model package. Runs 100% locally on any OS.