Analyze vision model failures
Company: Apple
Role: Machine Learning Engineer
Category: Machine Learning
Difficulty: medium
Interview Round: Onsite
For a computer vision product, discuss the following:
1. Explain the core machine learning fundamentals that matter most in vision work, including bias versus variance, overfitting, class imbalance, evaluation metrics, threshold selection, calibration, and data leakage.
2. Solve this probability question: 5% of images in a photo stream are truly blurry. A blur detector correctly flags 90% of blurry images and incorrectly flags 10% of sharp images. If an image is flagged, what is the probability that it is truly blurry?
3. A production vision model has started underperforming after launch. Describe a structured process to diagnose root causes and decide on fixes. Consider data drift, labeling quality, train-serving skew, preprocessing bugs, feature issues, model capacity, threshold problems, and offline-online metric mismatch.
Overview: This question evaluates proficiency in computer vision machine learning fundamentals (including bias–variance tradeoffs, overfitting, class imbalance, evaluation metrics, threshold selection, calibration, and data leakage), probabilistic reasoning for performance interpretation, and production model diagnosis and root-cause analysis.