Users can deploy their applications on Flek Server in 3 modes: standalone, local or on a remote server hosted in the cloud.

Scalable Architecture for ML

When running Flek Server in Standalone mode, programs run ML and interact with the FlekML engine directly. In Local and Remote modes applications connect to the ML engine via HTTP.

deployment_stack.png

The flexible architecture of the Platform and the way the 3 components – engine, server, and toolkit – work together, allow the data scientists to design, test and deploy their applications seamlessly.

From a programmer's point of view, the coding style stays the same. Applications developed using the 3 modes differ only by the configuration and connection defined at the start of the program. Therefore, developers can easily develop and test locally and then, with minimal change, deploy on the cloud for performance and scalability.