Conclusion

Going forward

This doc hopefully taught you many things. However, it barely scratched the surface of machine learning code development. To improve further, test new frameworks, don't put blindly your faith in some practices written by anyone. Experience can provide you with the critical understanding of what you are doing.

A data scientist interacts with business analysts, IT specialists of all forms and shapes. I've learned an awful lot by learning the basics of the jobs of all teams I've worked with: front-end developer, data engineers, system administrators... It provides a great feedback on how my code would be integrated within the company's architecture.

Ultimately, code practice is the byproduct of the coding culture of your teams. Sticking to this culture at first is always a good move. Then one can try to improve it bit by bit.

So read, learn and onward to better ML practices!

Many thanks

This doc would not have seen the light without the guidance of Christophe early in my career, when code development wasn't something I knew much of. Romain, Yann and Marc were hammered by questions, being almost test subjects. Sorry, I hope that this doc will be clearer that what I was muttering.