In my experience, I have found Jupyter Notebooks as a very useful tool for research and prototyping. On the other hand, if used incorrectly, interactive notebooks allow for bad coding habits which ultimately punishes productivity, research reproducibility and quality.

In this talk I presented common problems with research code, as well as practical advices how to improve the quality of the research codebase.

Link to presentation: Presentation
Link to code demos: GitHub

Version of this presentation, focused on software engineering