For data analysis, especially preliminary analysis, the existing ML and scidata packages make it trivial.
lower volume micro services, csv generation, low volume web APIs, and just about everything that is not performance critical, mostly because of the ease of programming and you can find a library for just about anything.