Chronon is an open-source, end-to-end capabilities platform designed to help machine learning (ML) teams build, deploy, manage, and monitor data pipelines for machine learning.
Chronon allows you to leverage all the data in your organization, including batch tables, event streams, and services, to drive AI/ML projects without the orchestration management typically required.
Key features:
- It consumes data from a variety of sources including event streams, DB table snapshots, change data streams, service endpoints, and warehouse tables modeled as slowly changing dimensions, fact tables, or dimension tables.
- It produces results in both online and offline contexts: online as a scalable, low-latency endpoint for serving features, and offline as Hive tables for generating training data.
- Real-time or batch precision: Set the results to either temporal or snapshot precision. Temporal precision updates feature values in real-time in an online context, producing point-in-time accurate features offline. Snapshot precision updates features once each day at midnight.
- Backfill your training set from raw data without waiting months to accumulate feature logs for model training.
- Leverage a powerful Python API: Data source type, freshness, and context are API-level abstractions comprised of intuitive SQL primitives such as group-by, join, and select, as well as powerful extensions.
- Automated feature monitoring: Automatically generate a monitoring pipeline to understand training data quality, measure training serving skew, and monitor feature drift.
Chronon is freely available on GitHub.
Must read:
