I've also noticed a trend of college graduates who in interview struggle with general software engineering practices and more fundamental coding skills and CS knowledge, but have knowledge and proficiency of ML. If you're a CS BA or MS, and don't plan to do an ML PhD, I'd suggest asking yourself if you want a data science job or a software engineering one. If the latter, remember to focus more on that. Surface knowledge of ML is a bonus, but the rest is more important.
My partner has ditched crosstraining into data science and ML because of how ridiculously 'ductape and string' the entire sub-indistry is. Tooling that is pre-2000 quality if it even exists, people with math skills yet only the most rudimentary ability in best practice around data storage and maintenance, coding, domain knowledge and understanding of how much bias they're introducing into results. Her view is its just a free for all of people who have little to no need to justify their output because everyone is waiting and hoping for the magic to happen.
Larger organizations (by size and longevity) tend to have staff that predate the data science and ML/AI hype train and have therefore worked out more robust tooling. Unfortunately the popularity of ML is driving a lot of adoption of trendy but not necessarily best practices.
There was a lot of this sort of thing with the web application boom as well. Lots and lots of decrying how JS libraries and frameworks were ruining the software industry.
I've not following that very much in a decade though. Did that whole area end up maturing?
I've noticed this myself but wasn't sure if it was just me. Almost every CV we get has a ML slant but the candidates struggle to write a simple SQL query.
From a pure numbers perspective 95% of jobs in our industry have nothing to do with any reasonable definition of machine learning. Even those applications of ML have a large amount of traditional CS going into acquiring, reformatting and storing large datasets.
That’s definitely something I’ve seen too. Most of the cvs I get through, especially from younger devs, talk about ML. I’ve got my work cut out for me keeping my existing codebase understandably without deliberately introducing opacity!
