ZenML

ZenML

ZenML is an extensible, open-source MLOps framework for building portable, production-ready machine learning pipelines. It provides a pipeline abstraction that is agnostic to the underlying infrastructure — the same pipeline code runs on a local laptop, Airflow, Kubeflow, Vertex AI, SageMaker, or Azure ML without changes.

Key Features

• Infrastructure agnostic: run pipelines locally or on 50+ integrations (Kubeflow, Airflow, Vertex, SageMaker, Azure ML)
• Stack concept: decouple ML code from infrastructure config (orchestrator + artifact store + model deployer)
• Artifact versioning: automatic tracking of all data artifacts and model versions
• Step caching: skip already-run steps for faster iterations
• Built-in integrations: MLflow, W&B, evidently, Seldon, BentoML, Great Expectations
• ZenML Pro for team dashboards and managed infrastructure

Quick Start

from zenml import pipeline, step

@step
def load_data() -> pd.DataFrame:
    return pd.read_csv("data.csv")

@step
def train_model(data: pd.DataFrame) -> Any:
    model = RandomForestClassifier()
    model.fit(data.drop("label", axis=1), data["label"])
    return model

@pipeline
def training_pipeline():
    data = load_data()
    train_model(data)

training_pipeline()

Install via ai-supply

npx ai-supply add zenml-mlops-framework

Curated mirror of the open-source ZenML (Apache-2.0). Get it from the source.