Legion is a resilient AI Runtime platform to run AI Learning models in production. It was build around the idea of having flexible unified environment for research and production as well as streamlined process of delivering those models into production, monitoring their performance, identifying possible mistakes and tracing them back to the source code.
Legion provides:
- Effortless out-of-the-box CI/CD process, including integrated model quality control with automated retraining and regression testing.
- Cloud-agnostic AI toolchains integrated with the most popular Sklearn, Tensorflow, Spark, R.
- Integrated feedback loop for the input and output capturing, A/B testing and traffic management support.
- Python 3.6
- Ansible
- Docker
- Kubernetes
- Jenkins
- Nexus
- Graphana
- Graphite
- Prometeus
- Flask
- NGinx
- Sklearn
- Tensorflow (not supported yet)
- Spark (not supported yet)
- R (not supported yet)
Please find it here CHANGELOG
If you are interested in contribution to Legion feel free to create a pull request or create an issue report. If you interested in hearing more about Legion implementation, existing case studies, questions or concerns feel free to contact us at Legion discussion group.