Data Science, Code Quality Hierarchy of Needs

We know that data science is started with analysis and often it's relatively unprepared to care about this. And when we moved from data science to machine learning, where you would need to create something and then ship it and maintain it on the long term, it became an issue. Could you explain what the hierarchy of needs in terms of like code quality are for data scientists and how they can inflame semi-l products? Yeah. I do think that anyone who can get proficient in statistics will be able to learn this concept.