Meta · Primly Community

Meta interview rejection post-mortem, what I'd change

hardware_hugo · 4 replies

got rejected after the onsite last month. not a great feeling but i want to write this while it's fresh because i made some choices i'd absolutely change.

The rounds I had: two coding rounds, one system design, one JEDI (behavioral). E5 SWE target.

Coding: i think i passed these. moved through both problems okay, talked through tradeoffs, got working solutions with time for follow-ups. the hard-medium pair was manageable.

System design: this is where i think i lost it. the prompt was designing a notification delivery system at scale. i went deep on the database schema and storage layer early. spent probably 20 minutes there. by the time i got to the actual delivery queue, fanout logic, and failure handling, we were running out of time. the interviewer was subtly trying to redirect me toward rate limiting and push vs pull delivery tradeoffs and i kept steamrolling back to schema.

what i should have done: sketch the end-to-end in 5 minutes, then ask the interviewer what to go deeper on. that's what meta interviewers actually want. they're checking whether you can see the whole picture before you zoom in.

JEDI (behavioral): i thought this went well but now i wonder. i used the same project story twice (different angles but same underlying story). probably should have spread across different experiences. meta JEDI rounds want to see range. "tell me about a time you had conflict with a coworker" and "tell me about a time you pushed back on a direction" should be different stories.

What i'd change: System design: breadth-first, then go deep where the interviewer points you JEDI: prep 6+ distinct experience stories, not 3 or 4 Ask the recruiter during prep which round types are in my specific panel so i know what to weight

going to take 6 months before re-applying. i think i can clear this loop, just need to prepare differently.

4 replies

qa_quinn

the breadth-first system design point is really important and applies across all big-tech loops, not just Meta. interviewers want to see that you know what matters before you commit 20 minutes to one piece. state the approach, align on what to explore, then go deep. it sounds simple but most candidates skip the alignment step.

market_realist

yeah, i was so heads-down on my prep materials that i kind of skipped practicing that opening 5-minute sketch. i'd done plenty of deep dives but not the 'here's the shape of the whole thing' part.

infra_ines

for what it's worth, notification systems at scale is one of those prompts that's genuinely subtle. the queue vs direct-push tradeoff at 100M users is not obvious. rate limiting per user vs per device is a good trap. don't feel bad for getting pulled into the wrong layer. most people do.

frontend_fran

the duplicate story thing in JEDI rounds gets people more often than they expect. i made a spreadsheet mapping each of my stories to each behavioral competency so i could see where i had coverage and where i was thin. helped a lot.