Directus Ml Ops

This project is a set of extensions allowing developers of the directus community to train, test and use machine learning model inside their existing directus.

Details

• For somebody new in machine learning, it is a little bit to setup a decent infrastructure for the use of his model. Some developers whould be obliged to export their directus data in a json file, then train a model using python to finally embed the model in their nodejs backend just for simple predictions. This is really tedious and time consuming
• The main library I'm using is brain.js (https://github.com/BrainJS/brain.js) but I'm planning to add tensorflow js
• I'm trying to build a set of extensions, so I have chosen a bundle
• If I have more time, I will build a complete pipeline so some models can be trained every day with a cron task and only the best one in accuracy with be shipped to production for customer use.

Screenshots

Watch the video on youtube

How to run ?

1. Make sure you have created a .env file for directus configurations

2. Make sure you have docker and docker-compose installed

3. Run docker-compose -f docker-compose.demo.yml up in the root directory. the app will run on port 8055

4. Try to login:

   • Admin email: [email protected]
   • Admin password: @Admin2019