← Back to experiences
[ OK ] Loaded —
[ INFO ]
$ cd
$ ls -lt
01
02
03
04
05
$ ls -lt
01
02
03
04
05
user@intervues:~/experiences/…$
Level: Staff-Level
Round: Onsite · Type: Multiple Types · Difficulty: 5/10 · Duration: 360 min · Interviewer: Unfriendly
Topics: Data Structures, Algorithms, System Design, Behavioral Questions
Location: San Francisco Bay Area
Interview date: 2026-01-15
Got offer: False
Interview Rounds Overview
I recently interviewed for an E6 INFRA role at Meta. I felt positive about all the interviews, but I was ultimately rejected after the hiring committee round with no specific feedback, which was frustrating.
In the phone screen, I solved the problems quickly. The coding questions are similar to leetcode 26 and 71
During the onsite coding round, I also solved the problems quickly, and I easily handled the follow-up questions about edge cases. The coding questions are similar to leetcode 408 & 133.
For the AI coding round, I had to solve the maximum subset problem. I completed the first two test cases, but the last two timed out. Completing the first two test cases is usually enough to pass, and I haven't seen anyone complete the third. I tried all possible backtracking optimizations without success. DP might have been an option, but the problem asked for printing the maximum subset instead of just its length, which makes DP less suitable due to potential memory issues.
For the first system design question on online auctions, I explained the key points like pub/sub pattern, SSE for real-time BID result notifications, and message queue partitioned by auctionId. Once the interviewer gathered these key points, there were no further issues. I also spent some time estimating QPS.
The second system design question was about location-based service search, a simplified version. The input was longitude/latitude and radius, and the output was the topK locations within the range. The location type wasn't specified, so it could be fixed locations like Yelp businesses or dynamic locations like nearby Uber drivers. The focus was on the search method. I suggested three options:
The interviewer acknowledged the first two solutions, so I focused on how to create a custom spatial index for searching, covering the principles of geohash and quadtree, trade-offs between dynamic and static location searches, search patterns and specific algorithms, memory estimation for quadtree, and database sharding choices (hybrid mode of geohash and quadtree). I spoke for about 20 minutes, and the interviewer seemed very satisfied, nodding in agreement.
The behavioral questions were standard, and there were many of them. I don't remember them all, but the discussion lasted about an hour.
After completing all the rounds, the recruiter was very positive and responded to my emails quickly, using many emojis, making me feel confident. However, after being sent to the hiring committee, I was rejected. The recruiter sent a template rejection email and stopped responding. No specific reason was given. I'm under a one-year cool-down period for INFRA positions, but I can still apply for other roles immediately.