Code

We need to be careful about the code quality as well as the reproducibility of our project results. Some tools/practices that I found useful to build a reproducible workflow for our projects:

• Project structure: Organize Python code like a PRO & Cookiecutter Data Science
• This is a comprehensive structure, you can obviously diverge from this slightly, but I strongly recommend following this structure or any other structure that you find useful for your projects, here is one more example for PyTorch projects.
• Version control for ML (like GitHub for ML): DVC
• Managing ML models: ModelDB
• Please use an appropriate name for each GitHub repo associated with your projects: follow-the-rule, may_be_this_is_also_ok, But_Never_Use_This, HaveSeenThese, orThisAndThat.
• Code documentation
• Code documentation