40f39d8f
i just failed a google interview because i couldnt solve a segment tree problem in 25 minutes. i have 8 years of experience building distributed systems at scale.
the interviewer was nice but the format is broken. nobody uses segment trees in production. nobody. the entire hiring process selects for people who are good at competitive programming puzzles, not people who can actually build and maintain software systems.
and before someone says "well if you practiced more" — i did 400 problems. i just happened to get one i hadnt seen. thats the problem. the question pool is infinite and the time is 25 minutes.
eecb2b4d
segment tree in 25 min is wild. even people who've done that exact problem before would struggle to implement it cleanly in 25 min. thats a broken interview, not a you problem
a7a075a2
counterpoint: LC is the least bad option. what's the alternative? take-homes that take 10 hours? "culture fit" interviews that are just vibes? at least LC is objective
9a17b543
the alternative is what stripe does: practical coding + bug squash + integration. tests real engineering skills. but its harder to scale so big companies default to LC because its cheap to administer
a42c02bc
400 problems and you cant solve a segment tree? sounds like you were grinding easy/mediums and avoiding the hard patterns. quality > quantity
501918cc
this guy just told you he builds distributed systems at scale and youre telling him to grind harder on toy problems. this is the exact mentality thats broken about this industry
6f5fd96d
i hire people. LC is how we filter 2000 applicants to 20 onsites. is it perfect? no. does it correlate with engineering ability? loosely. is there a better alternative at scale? i genuinely dont know
fe43888a
the real problem is headcount. if companies werent getting 2000 apps per role they wouldnt need these filters. fix the supply/demand and you fix the interview process
