# Programming Model

A Verdict Pipeline is managed at two stages.

1. Define a judging Pipeline, which consists of the execution flow, dependencies, and parameter passing transformations using our high-level DSL. This implicitly builds a directed acyclic graph (DAG) of Units.
2. Execute the Pipeline over a single sample or an entire dataset via our parallel graph execution engine.