Project RAIDAR

Script2Service Example

This repository is meant to showcase how to turn a python script in which an ML workflow is implemented, into a web service that allows exploiting this workflow from a third party component.

Pre-requisites:

1. docker
2. python
3. optionally, a virtual env in which to run the original script

What is the workflow exemplified here:

This is a simple ML workflow using scikit learn, that takes a collection of documents, does tokenization and vectorization on them, and then performs text classification using a Support Vector Machine.

The idea:

Is to take this script, learn how to extract from it the models needed for production, and build a web API for them

Steps:

1. Take a look at the Script directory. There you will find the baseline script, which we will convert to a web API. The script itself, and the pipeline implemented therein is not so amazing.

2. Checkout the branch "service" of this repo, to see how this is later implemented into a web API based on the FastAPI library.

The reason there are two branches is to allow for the service to be constructed step by step during a workshop.