Install required dependencies

Start training python mnist_classifier.py.

This will download MNIST dataset to the current directory, create mlruns folder for storing training experiments logs, checkpoints and starts training.

Run mlflow ui to open mlfow dashboard and track training history.

Run mlflow models serve -m mlruns/0/<run id>/artifacts/model -h 0.0.0.0 -p 8001 to deploy the trained model.

The deployed model can be used using CURL or the implemented client inside mnist_classifier_client.py.