Azure ML Pipeline

flowchart LR
DS["Datastore<br/>(ADLS Gen2)"] --> Prep[Prep step]
Prep --> Train["Train step<br/>(GPU compute)"]
Train --> Eval[Evaluate step]
Eval --> Reg{AUC ≥ baseline?}
Reg -- yes --> Model[Model registry]
Reg -- no --> Stop[Stop pipeline]
Model --> EP[Managed online endpoint]
EP --> Mon[Data drift + model monitor]
Mon --> Alert[Alert on drift]