I've failed a bunch of these interviews recently — huge amounts of code, all in the details, barely any algorithms. Not sure if I'm just writing too slowly or what, but it feels like they're hiring an AI.
The take-home was the same as one already discussed on the forum — pretty simple, just understand the context, aggregate some data, then filter.
The hiring manager round had a bit of a weak red flag? The HM said something like: I like candidates who are mission-driven like you, because then you can push them and they'll work hard.
VO (virtual onsite):
1st: BQ (behavioral) had one slightly unconventional follow-up — why do you want to switch jobs? After I answered that I wanted impact, a fast-growing company, one with a mission, they followed up with: isn't your current company like that too? So why are you leaving? (hmmm.....)
2nd: practical algorithm — they didn't say anything about what it would cover, just called it "solve a real-world problem." Completely impossible to prepare for.
At the start they showed me the format and response of 2 APIs — one for activity with pages, the other where you give it a user id and it returns the name and so on.
The first question right off the bat: call the paginated API to get the total count plus the activities on that page, then print out all the activities. You need to call the API page by page — the function was already wrapped and provided, you just pass in a page parameter, but you have to figure out the page numbers yourself.
The second question was to group all the activities into "clumps" — at first I didn't even know what that word meant... I asked what the output should look like, one interviewer said "whatever you like," so I said okay, I'll return a list of lists. After I finished writing it, the other interviewer said it needed to be a clump object with 3 attributes: activity list, type, and name. So I went and built a class, wrote the return, and printed it out. Then the "whatever you like" interviewer said that should be right, but the other one said a single activity shouldn't be in a list — it should just return the single activity directly — so I was forced to change it again. By the time I finished changing it, 50 minutes were gone.
The third question was to print everything out, but using name instead of id. I didn't have enough time — I wrote the single-activity case first, but forgot that I needed to take the user id and call the API again to get the name. The interviewer who'd said "whatever" told me to just walk through it at a high level, since they wanted to leave time for me to ask them questions... it ended at the 55-minute mark and we moved to Q&A.
Then about an hour later, the recruiter called and said they weren't moving forward — the coding notes weren't good. I've failed a bunch of these lately, where they give you a scenario and it's just a huge amount of code to write. I don't know what to do anymore — is there anywhere besides Big Tech that still tests algorithms?
Discussion
Loading comments…