Liad
This might not be SO easy task without previous knowledge, but a good practice for new comers of how to create a docker that supports machine learning on python. There...
I think it should be a separate repository even
I've created a separate repository. This task should be altered to be moving what we currently have here in the research folder, to the other repository.
We should disable the Codacy rule that enforces it. For some reason - they are not up to date...
I would also add [this source](https://speakerdeck.com/pycon2016/brett-slatkin-refactoring-python-why-and-how-to-restructure-your-code) for how to refactor python properly
And regarding the serialization method itself, json is quite slow. [ujson](https://pypi.org/project/ujson/) is already faster, and there are also [other alternatives in the benchmark competition](https://medium.com/@abrarsheikh/benchmarking-avro-and-fastavro-using-pytest-benchmark-tox-and-matplotlib-bd7a83964453)