Model Containerizer for K8s
- Build models directly into DevOps-ready container images for inference (using MLflow under the hood)
- Supports parallel builds in Kubernetes jobs, using Kaniko, no Docker socket required!
- Generates Open Model Interface compatible images that are multi-purpose and portable, they work on multiple platforms: KFServing and Modzy
- Try the test drive today, then deploy our Helm chart to your K8s cluster to use it for real
The fastest way to get started is to use the test drive functionality provided by Testfaster. Click on the "Launch Test Drive" button below (opens a new window).
Talk & Demo
Follow one of our tutorials to easily get started and see how Chassis works:
- Tech Support Our tech support is graciously hosted on Modzy's Discord server in the "Open Source" channels
- Install with Helm into a Kubernetes cluster
- Build a container image from your model
- Deploy to KFServing the built image
Chassis is developed by Modzy, a commercial ModelOps platform designed to run any kind of machine learning and artifical intelligence model in production, at scale, with enterprise grade security, governance, and compliance. The design of Chassis grew out of Modzy's internal research and development to provide a way to easily containerize MLflow models to publish into the Modzy catalog and to support all kinds of models, both present and future, with first-class support for emerging capabilities like drift detection, explainability, and adversarial defense.
A full list of contributors can be found here.