|
Steve Jorbs posted:I would tell them no even if it wasn’t illegal.
|
#
?
Dec 8, 2021 11:20
|
|
|
|
| # ? Apr 16, 2024 15:04 |
|
|
raminasi posted:ask for the request in writing, either you have a small chance to get them in Actual Trouble or you get to see some funny floundering
|
|
#
?
Dec 8, 2021 11:27
|
|
|
Mr SuperAwesome posted:lol big O poo poo is something i am also terrible at because i studied physics not compsci, so i usually handwave this poo poo away with "uhh IO or network is probably slower, yolo" but really dict lookups are o(1)? i did celestial mechanics and radio astronomy later and i only know the o notation poo poo for the extent id expect to come up in an interview, so same actually. and yeah, python dictionary is what is called a hash map, so the access time complexity is o(1) average and o(n) amortised worst case - which makes it very useful for adhoc caching
|
|
#
?
Dec 8, 2021 12:22
|
|
|
Mr SuperAwesome posted:how much do you reckon you can get away with in interviews by saying "yeah well i would aim for an XYZ approach (hashmaps, always hashmaps) but in my professional experience CPU/mem optimization has never been the bottleneck, only ABC" and elaborate on how you would solve various other bottlenecks if that doesn’t work with the specific interviewer it’s either a faang interview or something you should walk out from, because python is not the language for algo fencing 9 to 5
|
|
#
?
Dec 8, 2021 12:26
|
|
|
to the big o not knowers, I used to be like you. i taught myself to program from books and recorded lectures and never really got it. i got sick of understanding how to solve problems but not how that silly notation worked. just set aside some time and learn it. it's really not that hard. once it clicks you won't forget it because it's pretty basic to conceptualize constant is obvious n is a regular rear end for loop, grows as data grows n² is a loop in a loop so it as data grows search time grows much more log n, log, etc, just look at a graph. It's math just Google "big o notation quizzes" I promise it won't take you more than a few hours to figure out, it's really not that complicated
|
|
#
?
Dec 8, 2021 12:47
|
|
|
https://www.bigocheatsheet.com that's also a good resource for big o stuff.
|
|
#
?
Dec 8, 2021 13:12
|
|
|
i think the problem is less so that the notation is difficult to understand - if you work in python, there’s no real impetus to know it. dictionaries are the only polished data structure in the core language, and everything else is going to be behind 10 abstraction layers, meaning that 90% of actionable algorithmic performance problems boil down to identifying that something you have is quadratic at best
|
|
#
?
Dec 8, 2021 13:19
|
|
|
also hash tables is a kind of unfriendly case to start out with, since it involves assuming a good hash function exists (i.e. everything gets distributed randomishly), the amortization of resizing (so either a complicated distribution of the work or some ops sometimes being linear), a constant key size (i.e. the '1' here involves hashing the key you're indexing with, so if the key is large that should be counted), and arguably some potential bad constants (stuffing a cryptographic hash function into things gives you pretty certain randomness, but it'll be quite slow in practice). hash tables are a bit abused even in theory, where people will slot a O(1) into their algorithm without commenting on the hash function, and in theory of course even a cryptographic hash can cause unbounded collisions in bad cases.
|
|
#
?
Dec 8, 2021 13:31
|
|
|
raminasi posted:ask for the request in writing, either you have a small chance to get them in Actual Trouble or you get to see some funny floundering
|
|
#
?
Dec 8, 2021 13:53
|
|
|
of you need a good excuse say you use your inbox for reminders and your todo list. it’s just so you don’t forget 
|
|
#
?
Dec 8, 2021 13:54
|
|
|
the landau notation , which includes big o, was invented by mathematical analysist edmund landau to gently caress around w taylor series half a century before electronic computers were a gleam in turing and von neumanns mind
|
|
#
?
Dec 8, 2021 14:51
|
|
|
hr at most companies is filled with enough stupidity and incompetence that they would actually ask for illegal poo poo in writing even though preventing employees from suing is their job don't send your pay stubs unless you already have an offer letter in writing from them.
|
|
#
?
Dec 8, 2021 14:58
|
|
|
Cybernetic Vermin posted:also hash tables is a kind of unfriendly case to start out with, since it involves assuming a good hash function exists (i.e. everything gets distributed randomishly), the amortization of resizing (so either a complicated distribution of the work or some ops sometimes being linear), a constant key size (i.e. the '1' here involves hashing the key you're indexing with, so if the key is large that should be counted), and arguably some potential bad constants (stuffing a cryptographic hash function into things gives you pretty certain randomness, but it'll be quite slow in practice). “amortized” is probably the most abused word in computer science classes it’s like frictionless in first semester mechanics
|
|
#
?
Dec 8, 2021 15:02
|
|
|
i have never met anyone that cared that the time complexity of a disjoint set forest was ackshully O(n α(n)), it might as well be linear for all practical sizes. really what people are interested in when asking about big-o stuff is whether you actually understand how the algorithm you've written works beyond just "does it produce the right answer". - is it doing something to each element that takes a constant amount of time? that's O(n) - is it doing something to each element that takes a number of steps proportional to the depth of a tree? O(n log n) - is it doing something to each element, that takes a number of steps proportional to the total number of elements? O(n^2) - is it doing something for every possible combination of elements? O(2^n) (don't be here if you can help it)
|
|
#
?
Dec 8, 2021 15:08
|
|
|
is it theoretically able to run infinitely? best joke my friends made in CS at school was shufflesort, the most important algorithm taught in CS 666. shuffle array. check to see if it's sorted. if it is, you're done! if it's not, repeat.
|
|
#
?
Dec 8, 2021 16:03
|
|
|
In theory, shufflesort will terminate in finite time with probability one. In practice, you could be limited by sequences a PRNG can produce. Better use a hardware RNG.
|
|
#
?
Dec 8, 2021 16:05
|
|
|
silvergoose posted:is it theoretically able to run infinitely? you can optimize this to O(1) in quantumsort, which takes advantage of the "many worlds" interpretation of quantum mechanics. After shuffling, if the array is not sorted, destroy the universe (implementation left as an exercise for the reader). All remaining universes are ones in which the array was sorted successfully.
|
|
#
?
Dec 8, 2021 16:08
|
|
|
zombienietzsche posted:you can optimize this to O(1) in quantumsort, which takes advantage of the "many worlds" interpretation of quantum mechanics. After shuffling, if the array is not sorted, destroy the universe (implementation left as an exercise for the reader). All remaining universes are ones in which the array was sorted successfully. quantum bogosort and time loop logic were my favorite jokes in undergrad (actual “paper” https://web.archive.org/web/20090129114503/http://www.frc.ri.cmu.edu/users/hpm/project.archive/general.articles/1991/TempComp.html )
|
|
#
?
Dec 8, 2021 16:39
|
|
|
i actually implemented sleepsort in production once to try and shard some horrible job that was taking ages (for other stupid reasons) based on UUID across a ~day interval was quite happy with that one (sadly didn't solve the problem in the end, because the rest of our stack sucked)
|
|
#
?
Dec 8, 2021 16:50
|
|
|
silvergoose posted:is it theoretically able to run infinitely? how to tell if your brain is busted enough to post in yospos: https://www.youtube.com/watch?v=kPRA0W1kECg you enjoy this video
|
|
#
?
Dec 8, 2021 17:12
|
|
|
Real yosposters watch the extended version: https://www.youtube.com/watch?v=xoR-1KwQh2k Note: I am not a real yosposter.
|
|
#
?
Dec 8, 2021 17:21
|
|
|
i firmly believe this could be something you could show on sesame street and kids would be transfixed by it https://www.youtube.com/watch?v=ywWBy6J5gz8
|
|
#
?
Dec 8, 2021 17:30
|
|
|
PIZZA.BAT posted:i firmly believe this could be something you could show on sesame street and kids would be transfixed by it haven’t thought about this video, which kicks rear end, in a long time
|
|
#
?
Dec 8, 2021 20:55
|
|
|
Cybernetic Vermin posted:also hash tables is a kind of unfriendly case to start out with, since it involves assuming a good hash function exists (i.e. everything gets distributed randomishly), the amortization of resizing (so either a complicated distribution of the work or some ops sometimes being linear), a constant key size (i.e. the '1' here involves hashing the key you're indexing with, so if the key is large that should be counted), and arguably some potential bad constants (stuffing a cryptographic hash function into things gives you pretty certain randomness, but it'll be quite slow in practice). Early on in my career I learned that if you explain this to an interviewer who is just looking for O(1) as the answer that you won’t get the job.
|
|
#
?
Dec 8, 2021 22:15
|
|
|
tk posted:Early on in my career I learned that if you explain this to an interviewer who is just looking for O(1) as the answer that you won’t get the job. otoh for the best-paying job i've gotten i got into a shouting match with the interviewer about some tree balancing detail. (i do not advice this approach)
|
|
#
?
Dec 8, 2021 22:17
|
|
|
I’m pretty glad I didn’t get that job because I didn’t learn until much later that I don’t want to work at a place that doesn’t hire you for being right.
|
|
#
?
Dec 8, 2021 22:29
|
|
|
tk posted:Early on in my career I learned that if you explain this to an interviewer who is just looking for O(1) as the answer that you won’t get the job. what you should've learned is that you didn't want that job to begin with edit: tk posted:I’m pretty glad I didn’t get that job because I didn’t learn until much later that I don’t want to work at a place that doesn’t hire you for being right. lol yes this is right
|
|
#
?
Dec 8, 2021 22:29
|
|
|
PIZZA.BAT posted:i firmly believe this could be something you could show on sesame street and kids would be transfixed by it i watch it every time so i guess i am also transfixed by it
|
|
#
?
Dec 8, 2021 22:31
|
|
|
I’ve had almost every place I’ve talked to want way more information on where I’m currently at in interview processes with other companies than I ever have before. Desperate to get people to finish their processes. Seller’s market baby.
|
|
#
?
Dec 10, 2021 18:02
|
|
|
Iverron posted:I’ve had almost every place I’ve talked to want way more information on where I’m currently at in interview processes with other companies than I ever have before. Desperate to get people to finish their processes. Seller’s market baby. im still in two processes from September 3 and 5 respectively. i suspect neither will conclude before Christmas
|
|
#
?
Dec 10, 2021 19:06
|
|
|
cinci zoo sniper posted:im still in two processes from September 3 and 5 respectively. i suspect neither will conclude before Christmas Are you applying to government jobs?
|
|
#
?
Dec 10, 2021 19:38
|
|
|
i applied to 3 different tier-below-faangs 1st Dec and the processes seem fast as hell despite being xmas, talent screens within a week, 1st round within a week of talent screen for 2 outta the 3
|
|
#
?
Dec 10, 2021 19:55
|
|
|
i got the paid leetcode for a month thing to up my confidence too, they have tagged questions for a couple of the companies i'm going for and the questions actually seem fairly easy by leetcode standards which is nice i wonder how likely it is that you'll get a leetcode question tagged with $company in a real interview tho, seems too good to be true. (then again i ask candidates fizzbuzz in every interview i do, and still a good 30-50% fail, and like 20% have even heard of it, so who knows)
|
|
#
?
Dec 10, 2021 19:59
|
|
|
ultrafilter posted:Are you applying to government jobs? no, one had me on waitlist for a month and then actually came back and resumed the process. the other is just unusual in many other ways as well, but people seem cool and so does the job + comp is competitive, so im not minding the lengthy process any much actual government jobs here are much quicker. non-executive positions in latvian government take at most a month to receive an offer, and the one eu central government position that found me was ready to have me on board 3 weeks later
|
|
#
?
Dec 10, 2021 20:12
|
|
|
government salaries are lol here, altho the "working tax free for the EU" sounds kinda nice. altho i think you have to live/work in brussels for that one?
|
|
#
?
Dec 10, 2021 20:38
|
|
|
Mr SuperAwesome posted:government salaries are lol here, altho the "working tax free for the EU" sounds kinda nice. altho i think you have to live/work in brussels for that one? Brussels is pretty nice tho
|
|
#
?
Dec 10, 2021 20:47
|
|
|
after 11 years, i’m interviewing again. anyone get the change machine coding test where you make subclasses of coins and return objects of the right coins that make up the sent price?
|
|
#
?
Dec 10, 2021 21:00
|
|
|
i recommend you crack open a pure knapsack impl on matrices and do the objects as a finishing touch
|
|
#
?
Dec 10, 2021 21:01
|
|
|
Mr SuperAwesome posted:government salaries are lol here, altho the "working tax free for the EU" sounds kinda nice. altho i think you have to live/work in brussels for that one? eu offered me 320/day brutto for a non-senior data engineer position at an unnamed structural unit. with latvian government that would’ve been 53-64/day if you work for eu from brussels you pay no taxes, which for mid-senior specialists means 80-120k/year taken home. don’t look at rents around berlaymont
|
|
#
?
Dec 10, 2021 21:03
|
|
|
|
| # ? Apr 16, 2024 15:04 |
|
|
bob dobbs is dead posted:i recommend you crack open a pure knapsack impl on matrices and do the objects as a finishing touch i really messed up bad trying to understand what they wanted after being nervous when put on the spot, even thought it was so obvious. i went back through it and tidied it up and added errors for negative numbers and stuff, and made functions fileprivate that shouldn’t be exposed and sent an email of the code. hopefully it makes a better impression it was my 2nd interview and two hours long
|
|
#
?
Dec 10, 2021 21:07
|
|