Meta Interview Experiences

1. 🔹

   Screening (December 2024)

   This whole process actually started two months before the real onsite loop, with a short screening call in December. Two coding questions, both fairly standard, and the call wrapped up in about 45 minutes. I didn't hear anything for almost six weeks after that, and honestly assumed it hadn't gone anywhere until the recruiter emailed in late January to schedule the full onsite loop.

2. 🔹

   Onsite Round 1

   First real onsite round, two questions again, 35 minutes on the clock — tighter than I expected. First was simplifying a Unix-style directory path, which meant handling .., ., and repeated slashes using a stack. Got that one done in about 12 minutes. Second was finding the diameter of a binary tree, which needed a small recursive helper tracking height while also updating a running max diameter — I explained the two-birds-one-recursion idea before coding it, and finished with about five minutes left to walk through a test case.

3. 🔹

   Onsite Round 2 — Behavioral

   No coding this round, just one behavioral question after another — somewhere between 7 and 10 of them, which felt like a lot to fit into 35 minutes, so the pace was fast. I kept my answers tight and specific rather than long, since I could tell time was the real constraint here, not depth. A couple of the questions circled back to something I'd said earlier in the round, which meant staying consistent across answers actually mattered.

4. 🔹

   Onsite Round 3

   This round had the most follow-ups of the whole loop. Main question one was course-prerequisites, which I mapped to topological sort using Kahn's algorithm pretty quickly. Main question two was counting subarrays that sum to a target, solved with a running prefix sum and a hashmap. After I finished both, the interviewer kept adding small follow-ups — implement power(x, n) efficiently, then merge three sorted arrays with duplicates removed — almost like a rapid-fire round two. I got through all of it but felt noticeably more rushed by the end than in Round 1.

5. 🏗️

   Onsite Round 4 — System Design

   Last round of the loop, and it was system design: design a news-feed system, the kind of prompt I'd specifically prepped for. I started with clarifying questions about read vs. write volume, then sketched a fan-out-on-write approach for most users and fan-out-on-read for accounts with a huge number of followers, explaining why one size doesn't fit all here. We spent the last ten minutes on ranking — how to decide what shows up first — which I hadn't prepped as deeply, so my answer there was thinner than the rest of the round.

Key takeaway: All rounds ran under strict 35–40 minute limits — practice with timed mock interviews, sharpen edge-case handling under pressure, and stay adaptable across differing interviewer styles.

1. 📝

   Online Assessment

   The OA link showed up in my inbox on a Friday evening with a 5-day window to complete it, so I picked a quiet Sunday morning. It wasn't just coding — there were 4 or 5 separate stages, and one of them was a behavioral screening component mixed in with the technical stages, which I hadn't seen before at this step. I worked through the coding stages first since those had clear right/wrong answers, then went back for the behavioral part. Submitted it all in about two and a half hours, well under the time limit, and heard back about the onsite loop roughly ten days later.

2. 🔒 4 more rounds — continue this journey in the full kit

1. 💻

   Coding Interviews (Rounds 1–2)

   Two coding rounds back to back on the same day. First round was a subarray-sum variant — I'd seen the base pattern before, so recognizing the hashmap-prefix-sum trick took maybe two minutes, then the rest was just careful implementation. Second round mixed a k-closest-points question with a kth-largest-element question, both solvable with a heap, so I mentioned that connection out loud before writing any code, which the interviewer seemed to appreciate. Both rounds ended with a couple of minutes of buffer left, so I asked a question or two about the team back.

2. 🔒 2 more rounds — continue this journey in the full kit

1. 💻

   Day 1, Round 1 — Coding

   This loop was spread across two days, which I actually liked — less of a marathon in one sitting. Day 1 opened with an in-place merge-sorted-arrays problem. Simple on paper, but the "in-place" part meant I had to think carefully about merging from the back of the array forward instead of the front, to avoid overwriting values I still needed. Once I explained that idea out loud, the coding itself took under 15 minutes, and we spent the rest of the hour talking through a couple of edge cases — empty arrays, arrays of very different sizes.

2. 🔒 3 more rounds — continue this journey in the full kit