What’s it like to be a research scientist/data scientist in industry?
I’ll expand on my short answer, which is in the next paragraph.
It varies with the DNA of the organization. For example, the places I have been earn money in different ways and value different things.
- ETS (non-profit running tests like the GRE and TOEFL)
- Nuance (speech products, often on contract to undisclosed big company)
- Thomson Reuters (broad spectrum information provider)
- Digital Operatives (subcontractor to the security industrial complex)
- Facebook Applied AI. (trying to suppress “harmful content”)
- Facebook Linguistic Engineering (linguistic data and processes FTW)
- LivePerson (chatbot services and products for Fortune 500-ish clients)
- LexisNexis (information with a legal flavor, mostly for lawyers)
If you are a student now you are acquiring skills that will please and amaze people who are in business.
- Communication. Do as much as you can, to as many audiences as you can, orally and in writing.
- Evidence. There is great value in collecting evidence and using it to change your mind when you turn out to be wrong.
- Persistence. Dealing with the fact that the original plan didn’t work as expected, but the problem still needs solving.
Absent from the list of skills is any particular technical tool. If I were giving this talk in 1990, people would be asking whether they could keep using Prolog or Lisp in the commercial world, or in 2000 whether XML and XSLT were going to be important, or now, whether the company uses Keras, PyTorch or MxNet. These are/were all perfectly valid questions, but the answers change as quickly as anything else on the Internet, so don’t count on that kind of expertise to get you where you want to go.