Qwak Architecture

Overview

Qwak platform is a fully managed solution, which offers multiple deployment methods while keeping the maintenance overhead.
The supported deployment methods are:

• Qwak Cloud: In Qwak cloud deployment all compute and data services are run on Qwak Cloud.
  This allows quicker deployments in Qwak’s private and secure Cloud environment. There is no overhead, and you can keep your data encrypted.

• Qwak Hybrid Deployment: Keeps your data and models saved in your own cloud environment, while still providing a managed solution without the maintenance overhead.
  In order to create a hybrid environment, Qwak provides an CloudFormation (See example) to run on the requested AWS account, the CloudFormation creates a cross-account IAM role which gives Qwak the needed permissions for environment setup and management.

The Qwak environment can be installed in 2 ways:

• Default (Recommended):
  Deploy into a new VPC. The deployment CloudFormation creates all the necessary components to ensure everything works as expected.
  Existing VPC:
• Deploy Qwak resources in to an existing VPC environment along with existing resources within the VPC

Data is encrypted by default. So whether you choose Qwak cloud or the hybrid deployment, rest assured that your machine learning models and data remain private and secure at all times. For more information please read Qwak’s Data & Privacy document.

High Level Architecture

The Qwak architecture includes:

• Control Plane: Managed in Qwak’s cloud, handles non-sensitive metadata. It communicates with, and delegates sensitive operations (such as models and data operations), to the data plane.

• Data Plane: Hosts the model repository, Feature Store, Qwak Inference Lake, and handles the model and data operations (model building, deployment, feature creation, etc).
  The Data Plane uses the following services:

  • EKS Cluster to manage Qwak services
  • S3 for storing model output data
  • Elasticache to manage the online feature store