I mean you're not wrong; they built working software in Python, solved a problem, scaled it, and learned about the problem it solved. Once they understood it, they could monitor it. Once they had numbers, they could look into improving it, taking their learnings on the problem and picking a different language / architecture to solve it.