Getting Started

Cursus turns a pipeline graph (a DAG of step names) plus a JSON configuration into a complete, production-ready Amazon SageMaker pipeline — resolving the dependencies between steps, wiring their inputs/outputs, and generating the SageMaker step objects for you. You describe what the pipeline is; Cursus figures out how to build it.

This section takes you from pip install to a compiled pipeline in a few minutes.

• Installation
• Quickstart
• Core Concepts

The 30-second picture

There are three ways in, from highest-level to lowest:

Path	You provide	Best for
Pipeline catalog	a framework/task + a config JSON	starting from a proven, pre-built pipeline (44 shipped DAGs)
CLI	a DAG JSON + a config JSON	reproducible builds, CI, no Python glue
Python API	a PipelineDAG + configs	full programmatic control / embedding in your own code

All three converge on the same compiler, so they produce the same pipeline from the same inputs.

Fastest possible start

pip install cursus

# discover a pre-built pipeline and inspect what config it needs
cursus pipeline-catalog recommend --framework xgboost
cursus pipeline-catalog get-dag xgboost_complete_e2e

# compile a DAG + config into a SageMaker pipeline definition
cursus compile -d my_dag.json -c my_config.json -o pipeline.json

Then head to Installation for the install options and Quickstart for the full walkthrough (catalog, CLI, and Python API).