DEV Community

sunflowerseed
sunflowerseed

Posted on

How come job interviews turned in LeetCode familiarity?

There are some big companies, in which the interviewer told me secretly

"if you don't do tons of LeetCode, don't even think about getting into our company"

Another one I heard was

"if you can't give the most optimal solution in the phone interview (which was a medium hard LeetCode problem), don't even think about getting an onsite interview"

For some questions in LeetCode, it can be we work in the real world (especially as Front End development) for 10 years, and never use once. So getting into the company just means doing a lot of LeetCode problems. How can it make sense, or should we just take it "the way it is"?

Kind of like SAT: if you can't give a synonym of "abnegation" or "adumbrate", you just get a lower score and can't get into your dream school, that's it.

There is also somebody who said he

got warmed up by doing 300 Leetcode problems, and then he did 600 of them

On the other spectrum, there are also people who said LeetCode doesn't work, and that the interviewers only want to see how you behave in a getting stuck or difficult situation, and some say want to see how you cooperate with interviewer and discuss and reach a possible solution... I wonder how true it is, as it usually seemed the interview just pushed and pushed and say your code is 100% correct, and what is better, and when better is given, what is the most optimal.

Sometimes, the most optimal for people who haven't seen the problem before took them days or even months to reach it.

Top comments (3)

Collapse
 
ahferroin7 profile image
Austin S. Hemmelgarn

And someone who’s studied tons of Leetcode problems is more than likely a decent developer.

No, not really. LeetCode has the same issues that most similar 'coding challenge' systems do, as well as the general concept of coding challenges as usually implemented in interviews. In particular, they only really show that you can translate from already engineered solutions to real problems to code, which is not what actual software development is about (if it was, we wouldn't need humans to do it).

Take question #4 for example (Median of Two Sorted Arrays). I can throw together a valid solution that meets the time complexity requirement in maybe 5 minutes in 5 different languages that they support and at least 3 more that they don't without having to look at documentation. That doesn't mean I can actually do any useful development work in those languages, it just means I understand the algorithm they want well enough from the description that I can implement it in any language that I have a basic understanding of the required syntax.

Now, if you're just trying to test a person's ability to 'translate this complex text into an algorithm', then yes, it's passable (but not great). Beyond that though it tells you very little about their ability to do useful development work.

Collapse
 
sunflower profile image
sunflowerseed • Edited

true... sometimes I wonder why when interviewers told me my 5 or 10 minute coding is 100% accurate, but I didn't give them the "most optimal solution", and they don't hire me, and I went back to daily life, open up some app, and the program isn't even really running correctly, let alone "is it most optimal".

Collapse
 
sunflower profile image
sunflowerseed • Edited

some good developers don't really bother to practice tons of LeetCode, so the company could be weeding out many good programmers -- weeding out the true positives.

and some bad programmers can study leetcode, 300, 600 of them, and got into a company. The companies are hiring false positives.