Keter (keh-ter) is a highly-scalable lab environment for coronavirus data science. This project is informed by lessons learned from past B.Next projects.
The goals of this project:
- Support distributed computing using an distributed actor architecture.
- Agents can run on heterogeneous hardware and be started and stopped on demand.
- Support Kubernetes as the first class or preferred orchestration solution.
- Shared subsystem of models and datasets across all actors.
- Distributed model and data cache that is version controlled.
- Use machine learning techniques that build on state-of-the-art.
- User interfaces that are mostly or entirely static and can scale to massive amounts of users.
This project is pre-alpha.
A project of B.Next.
Licensed under Apache 2. See LICENSE file for details.