Blueprints

n the context of the KIOptiPack project, the IoT platform FastIoT is developed ad used as a core piece of the software technology-stack. This repository contains so-called Blueprints, which are essentially code templates to realize Machine-Learning use-cases with the FastIoT framework. The goal of Blueprints is provide a starting point for developers to quickly implement Machine-Learning use-cases with FastIoT.

FastIoT ist a micro-service-architecture and uses a Nats as a message broker to communicate between services. This architecture which introduces some peculiarities in the implementation of Machine-Learning use-cases, especially when it comes to the training of models. The Blueprints in this repository are designed to provide Blueprints as code templates to speed up the development of Machine-Learning use-cases with FastIoT, especially for transferring and preprocessing data, training, storing and serving the models.

• Database Blueprints
  • What are Database Blueprints?
  • MongoDB Database Blueprint
  • MariaDB Database Blueprint
• Data Processing Blueprints
  • What are Data Processing Blueprints?
  • Data Processing Blueprint
• ML Training Blueprints
  • What are ML Training Blueprints?
  • Pytorch Regression Blueprint (WandB)
  • Pytorch Regression Blueprint (MLflow)
• ML Serving Blueprints
  • What are ML Serving Blueprints?
  • Pytorch Regression Serving Blueprint (MLflow)
  • Pytorch Regression Serving Blueprint (WandB)
• ML Consuming Blueprints
  • What are ML Consuming Blueprints?
  • Pytorch Regression Consuming Blueprint