Skip to content

System designLevel 4

ML systems

Easy

Medium

Hard

Continue Read the guide View as list

Recommendation & rankingPicking the best 10 items out of a billion in less than 200 milliseconds.3 min Feature stores & MLOpsPreventing training data from accidentally leaking the future.3 min Training data pipelines: freshness, labels and leakageA training pipeline has exactly one job, produce data that looks like what the model will meet in production, and almost every way it fails is silent.9 min Vector search & embeddingsSearching by meaning instead of exact keywords using multi-dimensional math.3 min Fraud detection (velocity limits)Stopping credit card testing scripts by looking at the speed of actions, not just the action itself.3 min ML fraud detection (imbalanced classes)How to train an AI to find needles in a haystack of legitimate transactions.3 min Real-time leaderboardsWhy sorting a million rows in SQL every time someone scores a point will melt your database.3 min ML content moderation (human-in-the-loop)How platforms manage 10 million posts a day without hiring a million reviewers.3 min ML ads bidding (eCPM)How an algorithm decides which ad to show you in less than 100 milliseconds.3 min