Intermediate · Systems

Deployment

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. The process of moving a trained model into a production environment where it serves real users.

Technical Definition

The process of moving a trained model into a production environment where it serves real users.

How it works

Deployment involves packaging the model, exposing an API, configuring autoscaling, monitoring quality and latency, and planning for rollback. ML deployment differs from regular software deployment because model behavior depends on data, which can drift silently.

Related Concepts

• Data Drift — When the statistical distribution of inputs to a deployed model changes over time.
• Model Serving — Hosting a trained model behind an interface so applications can request predictions in real time.
• Inference Endpoint — A network address that accepts inputs and returns model predictions.