11

After finishing my PhD in Computer Science (machine learning) and after several years as a university lecturer, I decided to transition from academia to industry. As a result of that, I have been working one year for a company, doing applied research.

In my latest performance review, my supervisor pointed out that my only negative point was that I still had an academic mindset towards research, and that I should be able to get better at what he called "risk analysis". He defined risk analysis as the skill of assessing in advance the potential benefits/drawbacks for the company of a given method/technique/algorithm. The aim of that is to be able to rapidly discard methods that are supposed to not solve the company's problems without having to waste too much time on implementing them.

I was wondering whether any learning resource exists (book, online course/resources) that may help me to acquire such a skill. More generally, any recommended reading about computer science research in industry would be much appreciated.

  • 3
    This question appears to be off-topic because it not about academia; it is seeking resources about a specific skill for industry. – EnergyNumbers Jul 8 '14 at 13:57
  • 14
    @EnergyNumbers This site is for academics as well as anyone in or interested in research-related or research-adjacent fields which includes industry research. You can debate whether "risk analysis" is really a research skill or not, but the fact that it takes place in industry research and not academia does not make it off-topic. – ff524 Jul 8 '14 at 14:00
  • 1
    When I was in engineering industry years ago, non-PhD managers often had this opinion of engineers with PhDs. Does your supervisor have a PhD? If so, you could ask them how they developed this skill. If not, then your manager is just "pointing out the obvious." Don't sweat it; you'll develop this skill over time. – Mad Jack Jul 8 '14 at 15:30
  • 3
    @ff524: I'd argue that "risk analysis", as defined here, is also used in academia - in "real" research, you also need to assess as early as possible whether a given approach/question/whatever will be fruitful, beneficial, whatever... or impractical, useless, whatever else. So +1 to your comment. – Stephan Kolassa Jul 9 '14 at 6:49
  • 3
    I find it mildly amusing that your very question and approach reveals a purely academic mindset, too -- "please recommend some books that can help to dig into the issue deeper". I mean, no offense, this is a perfectly legit approach, and I tend to do the same. What really helps is just understand the basics: how your particular company makes money and how economics works in general. For starters, try to measure any solution in dollars. E.g., what is cheaper: to research how to automate a certain manual process or to outsource it to low-pay workers? – rg_software Jan 5 at 22:29
3

As a researcher in industry, let me first emphasize to you that "industry" is a much broader set of organizations and varies a lot more than most people realize. Thus, the answer to your question may depend quite a bit on the specifics of the company that you are at.

That said, however, in most cases a key animating idea for researchers in industry is a notion that often goes by a name like "customer focus": i.e., who will care about the results of this investigation?

Sometimes this is pretty straightforward (e.g., "if widget-making is 5% more efficient, then profit margins on widgets go up), but in other cases the relation is much more abstract or indirect (e.g., some prior work I did on potential programming languages for quantum computers that don't currently exist, but where the funder wanted to see how thinking about this might affect their goals for other research projects).

In all cases, however, you always have to be aware that somebody is paying the funds to support your salary and the salaries of your team, and they are giving you that money because they want some benefit to come from your research. This is true in academia as well, but professors are typically more insulated from it because their salaries are generally mostly supported by teaching---and even graduate students can be supported by TAships. If you're being paid to teach and expected to do research on the side, then it doesn't really matter what you're researching in the short term---but you will have more impact if you work on more important problems.

So, to the heart your question: how do you actually go about doing that?

I recommend starting by trying to be aware that at all times there is a "frontier" of research problems that you could be working on. Whenever you do a piece of work, you are choosing to prioritize one piece of that frontier over others. Notice this fact and ask yourself questions like:

  • Why did I choose this problem over those other ones?
  • How will solving this problem affect the other problems on my frontier?
  • Who else cares about the solution to this problem?

There can be lots of reasonable answers to these questions, but starting to ask them can be extremely helping in moving to a more conscious evaluation of your research choices and whether they're actually moving you in the direction that you want to go in your career.

For more thoughts and suggestions along these lines, I would recommend the following resources:

  1. "You and Your Research" by Richard Hamming, a long-time industrial researcher at Bell Labs.
  2. My own talk on "Surviving Life as a Researcher"
-3

Here are some links for you

this blog

books from Amazon

more books from Amazon

I am not interested in risk assessment but if I was, would sort the books by customer review, would see what the people who read them think, if possible would look inside the book.

Have to say though that I find strange you didn't do it yourself before posting the question.

  • 1
    Thank you for your answer. I, indeed, did several searches on Google, Amazon, etc. prior to asking the question. The issue with that approach is that the amount of published information is overwhelming, and therefore I was expecting somebody to point me in a more right direction. – Pablo Suau Jul 16 '14 at 10:42
  • 2
    This reads like more of an admonishment than an answer – benxyzzy May 1 at 9:03
  • this does not appear to be the "risk analysis" the question is referring to. – Marco Lübbecke Jul 8 at 22:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.