Machine Learning Engineer • Mid-Level
Machine Learning Engineer • Mid-Level
High QualityPractice Other Questions that Snap Asked
Continue with targeted prep from Snap's coding question bank and system design questions.
Interview Summary
I recently passed Snap's L4 Machine Learning Engineer onsite interview. The process covered ML fundamentals, binary search coding, recommendation system design, and applied ML, with a strong emphasis on video recommendation systems. After the initial debrief, the team added another system design round focused on detecting harmful content.
Interview Details
Round 1 — ML Fundamentals
This round covered a mix of core machine learning concepts and practical model evaluation questions. The interviewer asked me to name different optimization algorithms I was familiar with and discuss offline evaluation metrics. Normalized Entropy (NE) came up specifically, as it's a metric used at Snapchat.
Other questions included:
- Confusion Matrix: Identify the four outcomes in a binary classification confusion matrix: true positive, true negative, false positive, and false negative.
- Model Calibration: Explain model calibration and how to evaluate whether predicted probabilities are well calibrated.
- Overfitting: Discuss how to identify and address model overfitting.
Round 2 — Coding: Find the Peak in an Array
The coding question involved finding a peak element in an array with a specific
Unlock Full Details
Create a free account to unlock this interview experience report.