FAANG interviewing is purposely flawed
FAANG interviews stink on purpose. Because so many qualified candidates apply, FAANG interviewers have the luxury of saying “no” to lots of high quality people, and know there are more options.
For example, Amazon has a large candidate pool, because lots of engineers want to work there. The pay is competitive, and the reputation is good enough.
Amazon formerly encodes strict hiring, via the "Bar Raiser" process. The "Bar Raiser", is a special interviewer, who "sets the bar" by requiring that this candidate needs to be better than 50% of employees already in this role. When designing interview questions, Bar Raisers ask interviewers to predetermine "bar raising" answers for their question.
For LeetCode style questions, where a candidate needs to program a challenging algorithm on the spot, Meeting the Bar means finishing the question, but Exceeding the Bar, is finishing the question, and having time left over to answer extension questions. Extension questions are where follow-ups are asked, and the interview difficulty is ratcheted up. Oftentimes candidates know they got the question correct, and are confused when they don't get an offer. That's because not only do candidates need to get it right, candidates need to get it right so quickly that they have extra time left over to solve the extension, in order to get to the "Bar Raising" level.
This process invariably screens out people who are capable. By requiring exceptional results in the interview process, FAANG reduces their risk of bad hires.
Ssuch high standards exist because a bad hiring decision is very expensive. It can take months to years to detect and fire a bad hire. The cost of being wrong is higher than the cost of extra interviews and screening.
Non-FAANG companies can use this to their advantage, by picking up high quality candidates who don’t fit well into the very contrived mold of the LeetCode style programming interview. FAANG's hiring inefficiencies can be a competitive advantage for competitors, who can afford to be wrong, and fire quickly. Alternatively, better skill assessment interview practices than LeetCode could get the high quality candidates that FAANG has said No to.