I don't know why you think startups can afford false negatives. They absolutely CANNOT. One bad hire will kill a startup. Even spending the one hour to interview someone is a significant hit to productivity. It's the big FAANGs fo the world that can absorb poor engineers.

Also, for senior positions, I would expect candidates to not only clear most leetcode interviews, but to also have some insight from their experience. I expect them to be opinionated about areas of their experience, and have demonstrable success to back up their opinions. Only then can they come and lead and mentor a team of more junior contributors. A general guide line is, as a senior engineer, I would have hoped to have heard some insightful ideas or learned something from you in the course of your interview.

If you are clearning leetcode problems fairly easily, I wouldn't attribute it to a crowded market and "visa holders" being even better at leetcode (I don't even know what that means, do you think they go to better schools or something?)

I mean this is exactly why I am going to pretend to have less professional experience. These expectations have nothing to do with just doing more of the same work, but at a new company.

Of course I'm fairly certain it is something about me thats the problem - this round. I don't know what that is, the recruiters don't know what that is, the feedback looks just like bad luck as I've had wildly disparate results that weren't really only rejections, no different than the blogger in the article wrote.

Next round my solution is to come across as less impressive to lower the expectation and just join a team.

No, I don't think visa holders are inherently better and especially not all, it is not insensitive to acknowledge greater selective pressures from the countries they came from due to pure numbers alone, as well as desire to stay within this country in the short time period they have to find a sponsor if they get laid off. This can result in greater motivation to study more and faster, and you can see this corroborated by the self-certified number of leetcode problems people say they do. The mere presence of these motivations from some unknown number of people, distorts the market. Not really that hard or fantastic to say that.

I also said I was only clearing Leetcode easy problems easily. Which is a caliber that sometimes comes up in some interviews, with nothing harder presented. With harder in different interviews I was just starting to make the 30 min mark and other bars. It still sucks if you realize your first solution isn't good that there is no time to double back.

I think market now has moved on to leetcode DP/hard in ~30 mins.

I got this in my last phone screen at tier 2-3 shop

https://leetcode.com/problems/split-array-largest-sum/

DP?

and yes, I got good at the easy problems, which would be all that appears in some interviews. In other interviews I could tell or got feedback that I was close to the bar, which is about 30 minutes.

Or if I had an opportunity to finish a harder question later after the interview, I could tell that writing a working solution would have still taken me MUCH longer.

> DP?

Dynamic programming, for the problem I linked.

Oh maybe I am having bad luck

I got hit with another DP/Hard ,

https://leetcode.com/problems/word-ladder-ii/

at company called ' The Trade Desk' , which i believe is no where near Tier 1/2 tech companies. Exhaustive graph search ins't an acceptable solution.

what is an acceptable solution? what criteria does the acceptable solution solve such as computation time and space, or is that a factor at all?

Not sure. Thats what interviewer told me before he failed me. I am guessing he was expecting some sort of caching, memo soln.

Just did that one, didn't find the actual coding hard at all. But, just understanding what it wanted took a lot of time which would make me nervous, which would make me look stupid under the gun.

I don't think I could do it confidently with someone looking over my shoulder, clock ticking, and $200k in sitting in a suitcase.

> Just did that one, didn't find the actual coding hard at all.

what what the time complexity of your solution? I failed the interview cos n^m soln wasn't accepted. he said hes looking for logarithmic soln.

It could be improved yes. Earlier I noticed I only split the array into two and not into m pieces. So later went on a tangent on finding the best way to do that. But, I made the test case happy.

Also it says the length of the array is under 1000, so such an optimization is unnecessary.

The point though is that I probably couldn't do the basic solution with a gun to my head, much less an optimized one. One whiff of struggle and I'd panic and it'd be over at that point. I get very good reviews on my work under normal conditions.

Expecting an optimized, "original" work in a watched/timed environment is ridiculous imho. Interviewer is clueless and destined for failure.

I'll keep this link however. Next time a poor interviewer comes to me with similar I'll first state, I don't do 'swordfish' problems. If they insist, I'll challenge them to this one first. :D

ok this is what i got

   def splitArray(self, nums: List[int], m: int) -> int:
        @lru_cache(maxsize=None)
        def minSum(pos,count):
            currentSum = 0
            minSoFar = float("inf")
            for i in range(pos,len(nums)):
                currentSum+=nums[i]
                if count !=1:
                   next = minSum(i+1,count-1)
                   maxSplit = max(currentSum,next)
                   minSoFar = min(minSoFar,maxSplit)
            return currentSum if count ==1 else minSoFar

        return minSum(0,m)

I think putting that @lru_cache annotation is what makes this a Dynamic programming problem. If thats the case, i think its resonable interview question.