MLOps Demo

This repository contains demo code to showcase the complete pipeline for using MLOps

Project Organization

This Project is based on the cookiecutter data science project template. #cookiecutterdatascience

The demo code requires the following to work:

• AWS access key and secret key stored in the .aws/credentials file under the default namespace
• The AWS user should have access to Elastic Beanstalk, S3 and EC2 instance running MLFlow in it
• The raw dataset can be found in the UCI website
• Upload the dataset to S3 and configure the bucket name in the params.yaml file

AWS

• In the AWS configure a EC2 machine with port 8080 open and install mlflow and instantiate the server
• Update the params.yaml with the URL to the EC2 machine this will help to set the tracking uri for MLFlow

Local Machine

• After updating the params.yaml file with appropriate paths run the dvc repro command from the project root
• The result app will be deployed to your AWS Elastic Beanstalk environment