Most Common Interview Questions

[T3sl4co1l]

another interesting one to an interveiw i have seen,

Q: hold your hand / arm out straight and keep your fingers as still as you can.

it was a test to rule out people with hands too shaky to solder SMD parts

I'd fail that one due to arthritis but can do SMD soldering just fine. Why not ask to see examples of work instead?

19/20 applicants probably don't know what a soldering iron is.  What work?

If you've done SMT before, just mention it on your resume and bring it up in the interview.  This is why the interviewee can talk, too

[zapta]

another interesting one to an interveiw i have seen,

Q: hold your hand / arm out straight and keep your fingers as still as you can.

it was a test to rule out people with hands too shaky to solder SMD parts

Some jobs do require a steady hand

[zapta]

which suggests that the question is a good prima donnas filter. It's not about the answer, it's about the attitude.

If that's their idea of a crude personality test I'd rather not work there anyway, thanks. If their reaction to themselves being morons who don't know how to interview and couldn't be bothered to google some good questions is to think I'm a prima donna... It says more about them than me. With that kind of attitude (blame others for your own failings) it's usually best to avoid them.

which suggests that the question is a good prima donnas filter. It's not about the answer, it's about the attitude.

If that's their idea of a crude personality test I'd rather not work there anyway, thanks. If their reaction to themselves being morons who don't know how to interview and couldn't be bothered to google some good questions is to think I'm a prima donna... It says more about them than me. With that kind of attitude (blame others for your own failings) it's usually best to avoid them.

That's a convenient way to think of it. There is also the possibility that you are missing on good jobs because of dogma and inflexibly.

In my last job interview majority of the questions were open ended. How do you determine the temperature outside?, how many unemployment request are filed in the US every day?  What happens when you click on the Go button in a browser? How a computer process can communicate with another processes? It's like a Rorschach test that allows to guage your attitude, knowledge, and ability to think outside the box and handle situations and uncertainty.

[zapta]

Don't let your dogma costing you good jobs.

[sean0118]

I'd have to say one that immediately comes to mind was "If you were an animal, what animal would you be and why?" ...

I was asked that by a well known telecommunications company. I said dog, because I have a dog and it seems like she has a pretty good life. I don't think that was the answer they were looking for, but I didn't really want the job, just interview experience. Anyway, everyone else was giving lame answers like a cat because they were so independent or an eagle because they are so good at keeping an eye on things...     

[mtdoc]

Having interviewed many people over the years - from technical staff for a science lab to medical staff and physicians, you learn to identify candidates who are just tryng to give the answer they think you want. For any job requiring independent, creative thinking, those kind if candidates score poorly.

The point about the interview also being about the job candidate interviewing the employer is also on the money IME - at least for anything but the most menial type of work.

Hearing some of these sort of pop psychology questions engineers are being asked amazes me. I can't believe someone employing an engineer actually finds value in those kind of questions.

[Seekonk]

I'm out of the system now.  I remember having an interview that went well and was called back for a second one.  Thought they wanted to talk price.  Instead, I was led to a room and handed a psyc test. You know..... What would you like to do best on a sunny day:  A. Wear women's clothes, B. Kill your mother, C. Have tea with your cat.  To top it off, this was a timed test and I didn't bring my reading glasses.  I probably wanted to look like I could read small parts without assistance.  The usual test taking terror set in and squinting to read the questions got my eyes watering.  I could barely get through half the questions.  Never heard from them again.  They must have thought I was slow witted.

[German_EE]

I forget which company it was but I had a psychological test once and this one was all yes/no answers asking if I agreed with a particular statement. After thirty seconds someone asked what I was doing and they were quite surprised to find me going through the paper marking all the statements I agreed with first (I then planned to go back and mark the other statements as disagree). Nobody had done the test that way before.

One of the questions was "Do you believe in UFOs"? which I marked as "To be discussed". I then had to explain the difference between a UFO and flying saucers full of little green men to the interview panel. Hint: If you see something in the sky that you cannot recognize then, by definition, you have seen a UFO.

[photon]

True story. After a long and tiring day of interviews, an interviewer asked me what my greatest weakness was. I answered that I did not like people. I still received an offer, which shows that you can be an engineer and a sociopath.

[photon]

I didn't perform well TBH, probably because it was my second interview that day or just I wasn't so smart on stress, however when going back on the train I've managed to come to a result of 120bits.

Can someone do better?

Okay, you managed to nerd snipe me for the last hour, but after getting stuck at 160, I gave up and consulted the internet.  According to people who know the rules of chess better than I do, the number of board positions is ~4x10^40, or ~135 bits, before accounting for pawn promotion (which I forgot you could even do!).   Some estimates are apparently even as high as 10^52 (175 bits), if you account for promotions.  How the heck do you get 120?

You can use probabilites to reduce the number of bits required to encode an outcome. So, in the example you give of 4*(10^40) =~ 2^135 outcomes, you will need 135 bits only if each of the outcomes is equiprobable. However, if this is not the case, then you can reduce the number of bits required to the entropy of the source. This is Shannon's first theorem and the basis for all lossless compression. 

[helius]

True story. After a long and tiring day of interviews, an interviewer asked me what my greatest weakness was. I answered that I did not like people. I still received an offer, which shows that you can be an engineer and a sociopath.

The word for that is "introverted". Sociopathy is a lack of conscience, emotional manipulation, fabricating exactly what others want to hear, and very easy lying. Your boss and your congressman are probably sociopaths.

[photon]

True story. After a long and tiring day of interviews, an interviewer asked me what my greatest weakness was. I answered that I did not like people. I still received an offer, which shows that you can be an engineer and a sociopath.

The word for that is "introverted". Sociopathy is a lack of conscience, emotional manipulation, fabricating exactly what others want to hear, and very easy lying. Your boss and your congressman are probably sociopaths.

When I was at school I had a professor who was at Michigan with Ted Kaczynski where they both did a PhD. When I told my wife this she really liked to kid me that she was going to turn me in as the Unabomber. I think she was joking. BTW, I am also joking.

[mikeselectricstuff]

I had a very interesting interview in Imagination Technologies when I had to solve how to save a game of chess using the minimum possible amount of bits in the memory.

I wonder if they ware also looking for a more complete answer - e.g. there is no point squeezing the last few bits if it needs a ton of extra code. The wording of the question could be taken to include memory used by the encode/decode process.
For a software company, a good answer ought to be something that is simple, easy to understand and easy to show it works, even if it isn't necessarily the most efficient.
Maybe all they were doing was looking for someone to dig further into their incomplete specification of requirements.

[bookaboo]

True story. After a long and tiring day of interviews, an interviewer asked me what my greatest weakness was. I answered that I did not like people. I still received an offer, which shows that you can be an engineer and a sociopath.

Language maybe nsfw

[AndyC_772]

For the chess game question, I'd be looking for the candidate to talk through some considerations, even if there's no definitive conclusion. Things like:

- how many different kinds of piece are there, and therefore, how many bits are required to encode them all?

- given that the number of each piece at the start of the game is fixed, is it more efficient to consider an indexed solution? (ie. record the position of each piece in terms of its numbered square, rather than encode each square individually).

- in the latter case, how do you cope with pawn promotion? Is it necessary to add extra bits for this, or is it OK not to be able to cope with (say) multiple queens, if the memory saving thus achieved is worth it?

- what's more important, the total memory usage of 1 game, or the average memory usage over a large number of games?

- is the code space included in the total? How about an arbitrarily large look-up table? Or at least, a custom LZW entropy coding table?

- given that all identical pieces are interchangeable, is there a saving to be had in the indexed case by sorting the pawns by some criterion before compressing the table?

[zapta]

For the chess state, ignoring history information, my *theoretical* approached would be along these lines, define an order among all the valid states and assign each a sequential number. This will be the minimum number of bits assuming even distribution of states and encoding time is O(1) because the number of states is fixed.

[helius]

for a fixed input size, any algorithm is O(1) 

[zapta]

for a fixed input size, any algorithm is O(1) 

Optimal encoding, optimal time complexity, problem solved, when do I start the job?

[HackedFridgeMagnet]

Optimal encoding, optimal time complexity, problem solved, when do I start the job?

When you work out how many valid states there are. 

[zapta]

Optimal encoding, optimal time complexity, problem solved, when do I start the job?

When you work out how many valid states there are. 

O(1),  my professor said that constants do no matter.

[KJDS]

When interviewing people I want to know if they have common sense and a sense of humour as much as their technical competence.

[Tallie]

When interviewing people I want to know if they have common sense and a sense of humour as much as their technical competence.

Right. Some things you can learn, some things you can't...

[sean0118]

Out of interest how do people answer these types of questions:

'How would peers/family/colleagues describe you?'


The two phrases that leap into my mind are 'Very good thank you' and 'Why don't you ask them yourself?'.

But I'm guessing those as responses wouldn't help land a job...   

[KJDS]

Out of interest how do people answer these types of questions:

'How would peers/family/colleagues describe you?'


The two phrases that leap into my mind are 'Very good thank you' and 'Why don't you ask them yourself?'.

But I'm guessing those as responses wouldn't help land a job...   

"Technically brilliant, with a great sense of humour, lots of common sense and boundless energy, but not very good at answering lazy interview questions"

[glynd]

I didn't perform well TBH, probably because it was my second interview that day or just I wasn't so smart on stress, however when going back on the train I've managed to come to a result of 120bits.

Can someone do better?

Okay, you managed to nerd snipe me for the last hour, but after getting stuck at 160, I gave up and consulted the internet.  According to people who know the rules of chess better than I do, the number of board positions is ~4x10^40, or ~135 bits, before accounting for pawn promotion (which I forgot you could even do!).   Some estimates are apparently even as high as 10^52 (175 bits), if you account for promotions.  How the heck do you get 120?

You can use probabilites to reduce the number of bits required to encode an outcome. So, in the example you give of 4*(10^40) =~ 2^135 outcomes, you will need 135 bits only if each of the outcomes is equiprobable. However, if this is not the case, then you can reduce the number of bits required to the entropy of the source. This is Shannon's first theorem and the basis for all lossless compression.

Just having a play, I figure that you can encode the whole lot in 192 bits (including multiple pieces the same)
Using an initial 64bit bitmap to list the occupied spaces, and then 4 bits to represent the pieces used in turn.

Managed to knock off another 30 or so bits by playing around with some RLE ideas, but that's as low as I got in a quick play