Question

Facebook | Meta | SW/ML| Interview Experiences | Set-2 | 2022

Answer 1

Situation
A week before my virtual onsite interview, an hire freeze announcement was on the media. However, my full day interview was conducted as scheduled. All the interviewers acted like nothing happened at Meta, they did their best interviewing me, and so did I.

Two sessions of coding interviews
All the problems were slight as well as decent variants of the facebook tagged problems.
I practiced top 50 facebook problems (https://leetcode.com/problem-list/top-facebook-questions/) + additional 30 problems from facebook tagged, sorted by frequency in last 6 months (https://leetcode.com/company/facebook/). Out of 6 problems including a phone screening, I could solve 5 problems and didn't solve 1 problem:

String / Hash Map / Stack + Monotonous Stack / Queue / Heap ... 3 problems
Trees / N-arys / Graphs / DFS / BFS / Topological Sort ... 2 problems
Sliding Windows / Two-pointer problems ... 1 problem
One session of ML system design interview
As expected, designing a recommender system was asked. Although a given problem per se didn't mention a recommender system, but you could drive it by asking clarification questions.

I've mainly prepared with the popular "Gr** the ML Interview" and "ML System Design". They were helpful to drive a given one sentence problem into a ML framing approach. After that template approach, this blog at Meta was super helpful: https://ai.facebook.com/blog/powered-by-ai-instagrams-explore-recommender-system/

Two sessions of behavioral interviews
Situational questions asking for more concerte examples. Similar questions from two sessions.

Lesson / Next
It was good opportutnity going through SW/ML onsite interview with a focused one month preparation. Given a current situation at Meta, it seems wise to prepare for other interviews targetting elsewhere.