Luigi logo

Luigi

Luigi is an open-source Python package from Spotify for building complex pipelines of batch jobs, handling dependency resolution, workflow management and visualization.

PythonOpen SourceBatch ProcessingETLHadoop
Stars
18,783
Forks
2,470
Contributors
642
Latest
3.x (PyPI)

About Luigi

Luigi was built at Spotify and open-sourced in 2012 to stitch together long-running batch processes such as Hadoop jobs, database dumps and machine learning steps. Pipelines are plain Python: each Task declares the tasks it requires, the output Target it produces and how to run. Luigi resolves the dependency graph, skips tasks whose outputs already exist, and ships a lightweight central scheduler with a web UI for visualizing progress. It has no built-in cron-style scheduler, so it is typically triggered by cron or another orchestrator.

Key Features

Tasks defined as Python classes with requires(), output() and run()
Automatic dependency resolution across the task graph
Idempotent execution: tasks whose outputs exist are skipped
Built-in targets for local files, HDFS, S3, databases and more
Central scheduler (luigid) with a web UI to visualize the dependency graph
Integrations for Hadoop, Spark, Hive, BigQuery and Redshift
Parameterized tasks and date-based pipelines

Pros

Very small learning curve: plain Python with few concepts
File/target-based idempotency makes reruns safe
Battle-tested at Spotify on large batch workloads
Lightweight to run: no database required for the scheduler

Cons

No built-in time-based scheduler; relies on cron or external triggers
Limited UI and observability compared to modern orchestrators
Scaling is limited; workers do not distribute tasks across a cluster on their own
Development pace has slowed compared to newer tools

Tool Information

Language
Python
License
Apache-2.0
First release
2012
Latest version
3.x (PyPI)
Maintainer
Spotify / community

Used By

SpotifyFoursquareStripeAsanaSeatGeek

Compare Luigi with other tools

See how Luigi stacks up against other workflow orchestration engines, feature by feature.