Machine Learning Engineer · Full Journey · Multiple Types

Meta — Machine Learning Engineer ✅ Passed

Level: Senior-Level

Round: Full Journey · Type: Multiple Types · Difficulty: 6/10 · Duration: 240 min · Interviewer: Friendly

Topics: AI Coding, Recommender System Design, Binary Tree Traversal, Stack, Behavioral Questions

Location: San Francisco Bay Area

Interview date: 2026-01-10

Summary

Interview Rounds Overview

• Round 1: AI Coding
• Round 2: Machine Learning System Design (MLSD)
• Round 3: Coding
• Round 4: Behavioral

Details

I had a virtual onsite interview with Meta for a Machine Learning Engineer role.

Round 1: AI Coding

The coding was based on modifying code to pass unit tests in an IDE environment. The problem was a maze. Here's a breakdown of the questions:

• Question 1: Modify the code to keep the start and end points of the path as their original values (not '*'). I added an if-else statement to check if the current position was the start or end.
• Question 2: Debug a BFS implementation. The bug was the missing visited set, so I added it.
• Question 3: Add directional constraints to the path, where '>' allows movement from left to right and '<' from right to left. I added a check to the move method.
• Question 4: Implement a key-and-door problem, using a bit mask in the visited set to store the key state.
• Question 5: Add bombs to the path that destroy walls within a radius of 2. I added a bomb state to the visited set and used AI to generate the code, which passed the tests.

Round 2: Machine Learning System Design

The question was about designing a recommender system. I don't recall the specifics of what was being recommended, but the discussion focused on the overall architecture.

Round 3: Coding

• Given a binary tree, output the leftmost nodes from bottom to top and the rightmost nodes from top to bottom.
• Implement a calculator that only handles addition and multiplication. I provided two solutions: one using split and another using a stack. The stack solution required O(1) space complexity, and I had some minor bugs in my implementation.

Round 4: Behavioral

I felt this round was my weakest. I mentioned a project from 2-3 years ago, and the interviewer focused almost entirely on that project. I was asked about project planning, prioritization, collaboration, and conflict resolution. Since I didn't remember all the details clearly, I struggled a bit, but I appreciate the interviewer's leniency.