The Legion is an open source project dedicated to simplify automation of AI/ML products life cycle for small and large-scale enterprise systems.
Overview
- Modular Architecture
- Separate components for different AI product life cycle phases: training, packaging and deployment
- RESTful APIs and SDKs for all components to build extensions and plugins for external systems
- Command line interface to interact with system components
- Plugins for IDEs, Workflow Engines, CICD Engines
- ML experiments tracking component
- Logging, monitoring and alerting components
- Multicloud
- Core system components deployment in K8S with Helm Charts
- Deployment automation for major clouds: AWS, Azure, GCP with Terraform
- Security
- Single sign-on (SSO) with OAuth2 or SAML2
- System API endpoints secured with JWT tokens
- Credentials manager for connectors to external systems
- Multiple AI/ML Toolchains
- All major ML toolchains Scikit-learn, Keras, Tensorflow, PyTorch, H2O and others are supported
- Integrated with MLFlow open source project
Documentation
Changelog
Please find it here CHANGELOG
Backlog and Kanban Board
License
Contribution
If you are interested in contribution to project feel free to create an issue in backlog, create a pull request.