Any of you ever compete a case interview? Or other technical interview? Advice?

[TopHatFox]

Turns out my first interview this Wed is not simply fit (cake walk), but also involves me solving a real business problem using math and logic.

Good news is I learned that yesterday and started studying yesterday as opposed to Tuesday night. I found a neat book called Case In Point.

Any of you ever completed a case interview? Or ever completed a technical interview in general? What's your best advice?

[Tyler9000]

Just remember it's not just about the final answer but also about how you get there. If you don't know an answer, don't BS because they'll see through that immediately. Just stay calm and confident and talk through how you would go about finding the answer.

For complicated problems, a good technique is to ask questions along the way. Explore the constraints and assumptions. It shows that you're open-minded and a good communicator, and will score points over those who just take the football and run without really thinking about the problem space.

[IlliniDave]

I had one of those coming out of college. It was also a team thing. Really, it had nothing to do with technical acumen. It was a psychology test--rats in a maze--to see who behaved the way they wanted "under pressure". As soon as I figured that out (about 3 minutes in), and realized they had been disingenuous about the whole thing, it made me angry and I completely lost interest in that company on the spot. I'm sure that they lost interest in me about the same time.

So my advice is to beware, and understand it's quite possible they'll be looking at something other than (or at least in addition to) your technical acumen.

[NPV]

If you are interviewing for management consulting jobs, Google Victor Cheng case interview guide. Tons of useful info.

[akratic]

My advice is to order your interviews such that the companies you care least about interview you first.

Interviewing is a learned skill and you get better at it over time.

My favorite trick after they're done asking puzzles is to take control of the interview and steer the conversation, either towards the part of your background or resume that looks best, or with questions about their company. If you let the interviewer steer things and force them fill up the rest of the hour, you can get stuck with some retarded questions like "what would you say is your greatest weakness". It's much better if, for example, you have prepared questions that show you did your research about their company but want to learn more.

[jacob]

In [professional] science, the rule of thumb is that in order to perform well during the question-section of a conference talk (usually 1hr seminars have 5-10 minutes reserved for questions from the audience) you need to have read about 600% of what your talk covers. Here 100% would be reserved for your subject of expertise (all of it). The other 500% is considered sufficient to understand dumb questions from the audience as in "I know what you don't know (i.e. I understand why you're missing the point) and here's how I can explain what I know in terms that you understand". IOW, in order to communicate a significant amount of effort needs to be dedicated towards understanding where uninformed people are coming from. This is so that you can communicate with them and this is a good thing!

Case/technical interviews are kinda similar but they often completely miss the point.

For example, when I first tried breaking into Wall Street in 2008, the standard recommendation [for fallen scientists] was to spend 25% of the study time on memorizing the answers to brain teasers and practising sudoku. Seriously! A quarter of the time was to be spent on stuff that had zero applications beyond fooling an interviewer into believing that you were super-intelligent when in reality you had just memorized the answer to 150 easy brain teasers. And because of that if you couldn't come up with an answer within 60 seconds, the interviewer would figure that you were stupid. For example, I failed the fizz-buzz test during an interview because I never heard about it before and because I basically suck at thinking on my feet (I have a severe case of staircase wit). That's fair if the job would be to think on my feet. However, I can obviously nail that silly problem given 10 minutes or so in front of an actual screen/compiler rather than on a phone or on a whiteboard.

Today, the answer I would have given would be something like "Ahhh, the fizzbuzz test. Sure, but maybe we should talk about something more interesting." (Of course, I have FU money so I can get away with that ;-P ) ... but this speaks to the 600% rule.

Which means ... it pays off to study/understand how your interviewer works/thinks.

A quick entry to knowing more than the interviewer knows or cares to reveal that they know is reading the Personal MBA book. I think that this absolutely has to be backed by a personal project to lend credibility. I know this takes time, but I think it's time well spent when it comes to interviews.

Otherwise, I think a case/technical interview is just a re-run of college exam pop-quiz. If anyone were to ask me, I'd figure they're looking for an entry level cog position.

[Dragline]

I conduct a lot of interviews -- probably over 1000 in the past 20 years -- and have come to the conclusion that these "interview devices" of games or trick questions are more designed to fool the people doing the interviewing into believing that they can really reduce the uncertainty of the process. I don't know of any study that shows that you really can, as much of the uncertainty arises AFTER the person takes the job.

I have read articles that suggest that most companies hiring pools of people would be better off relying on the resumes/cvs alone or some other more or less mechanical process, because most people are actually quite bad at evaluating interviewees and make judgments based on superficial impressions.

[jacob]

@Dragline - Actually, employers would be best off relying on IQ testing their applications. This is probably why they aren't allowed to do so. The second-best option is to consider past historical performance(*). Interview performance ranks at the bottom in terms of predictive quality for job performance.

(*) Unfortunately, most recent grads have none.

I'm pretty sure that some general rule for "human systems" can be derived from this.---Although on second thought, I might be too 21st century blind.

[BRUTE]

brute agrees with Dragline - interviews are designed (evolved?) to lie to the interviewers and trick them into thinking they're not guessing.

jacob: if only IQ wasn't an easily trainable measure of anything actually relevant to anything besides the very specific skills tested for. brute knows people that trained IQ tests for fun and ended up maxing out all the tests they could get their hands on in 1/3-1/2 of the time provided.