Meta E5 interview

My interview process for the Software Engineer - ML role at Meta involved a phone screen and a virtual onsite, with a behavioral round still pending.

Phone Screen

In the phone screen, I was given two coding problems. The first was Subarray Sum Equals K, and the second was Palindrome Permutation.

Virtual Onsite

Coding Round 1

This round had two more coding challenges. I was asked to count the number of monotonic sequences in an array, and the second problem was K Closest Points to Origin, for which I implemented a heap-based solution.

Coding Round 2

The second coding round also consisted of two problems. The first was Valid Number, but specifically without the exponent use case. I initially missed the test case "05" which should return false, but the interviewer pointed it out, and I corrected my code. The second problem was a variation of Simplify Path. In this variation, I was given a current path and a relative path, and my task was to simplify the target path and then navigate from the current to that target. I solved this using a stack.

ML System Design

There was some initial miscommunication regarding this round, as my recruiter had mentioned it would be a general system design problem, but I received a core ML question instead. I was asked to design an ML pipeline for bot post detection. I went over all specifics including data preprocessing, feature engineering, training, testing, and evaluation strategies, considering both online and offline scenarios. I believe I had answers to most of the questions that came up, although the interviewer's reaction was neutral.

I'm now awaiting the behavioral interview. I'm a bit concerned if missing the "05" test case initially will negatively impact my coding round evaluation.