System Design
ML System Design Interviews
A practical ML system design course - a repeatable interview framework, an ML toolbox, and ten full case studies (news feed, content moderation, video recommendations, visual search, ad-click prediction and more) worked end to end.
ML System Design Interviews
Machine-learning system design interviews test a specific skill: can you take a vague product goal ("recommend videos", "moderate content", "predict ad clicks") and turn it into a defensible ML system - the objective, the data and features, the model progression, the evaluation, and the serving architecture - while talking through the trade-offs out loud.
This course trains exactly that. You start with a repeatable interview framework and an ML toolbox of the concepts that come up again and again (two-stage ranking, feature stores, offline vs online metrics, A/B testing, drift, bias and the cold-start problem). Then you work through ten complete case studies, each taken end to end:
- Personalized News Feed, Content Moderation, Video Recommendations, Visual Search
- Ad-Click Prediction, People You May Know, Event Recommendations, Video Search
- Street View Blurring, Similar Listings
Every case study follows the same arc - problem and discovery, ML framing and data, model development, evaluation and serving, and a summary with variations and interview tips - so the framework becomes second nature. The course closes with cheat sheets you can review the day before an interview.
The first two modules are free to preview. The full course is a one-time purchase with lifetime access.
What you will learn
- A repeatable framework for any ML system design interview question
- The ML toolbox: two-stage ranking, feature stores, candidate generation and ranking
- How to frame a product goal as an ML objective, with the right data and features
- Model progression (logistic regression to gradient-boosted trees to neural nets) and when each fits
- Offline vs online evaluation (AUC, LogLoss, NDCG, CTR), A/B testing, and drift detection
- Production serving architectures with sub-200ms latency, caching, and monitoring
- Handling bias, fairness, feedback loops, the cold-start problem, and class imbalance
- Ten full case studies worked end to end - from news feed to ad-click prediction
Prerequisites
- Working knowledge of machine-learning fundamentals (models, training, evaluation)
Course outline
Module 1: Welcome
Welcome - 1 topics.
Module 2: The ML Interview Framework
The ML Interview Framework - 7 topics.
- How ML System Design Interviews Work 6 min read
- Overview: The Five-Step Framework 6 min read
- Step 1: The Detective Phase - Clarify Requirements 9 min read Pro lesson
- Step 2: The Framing Phase - Frame as an ML Problem 8 min read Pro lesson
- Step 3: The Data Phase - Data Preparation & Feature Engineering 9 min read Pro lesson
- Step 4: The Brain Phase - Model Development & Architecture 8 min read Pro lesson
- Step 5: The Ship-It Phase - Evaluation, Deployment & Monitoring 11 min read Pro lesson
Module 3: The ML Toolbox
The ML Toolbox - 10 topics.
- Embeddings and Vector Spaces 12 min read
- Feature Engineering Patterns 14 min read
- Ranking and Retrieval 15 min read Pro lesson
- Classification and Regression 16 min read Pro lesson
- Neural Networks Simplified 17 min read Pro lesson
- Data Pipelines and Processing 13 min read Pro lesson
- Evaluation Metrics Deep Dive 15 min read Pro lesson
- Serving Infrastructure 13 min read Pro lesson
- Monitoring and Retraining 12 min read Pro lesson
- Ethics and Fairness in ML Systems 13 min read Pro lesson
Module 4: Case Study - Personalized News Feed
Case Study - Personalized News Feed - 5 topics.
- Personalized News Feed: Problem & Discovery 7 min read
- ML Framing & Data 10 min read
- Model Development 10 min read
- Evaluation & Serving 8 min read
- Summary & Variations 9 min read
Module 5: Case Study - Content Moderation
Case Study - Content Moderation - 5 topics.
- Harmful Content Detection System Design 7 min read
- Harmful Content Detection System Design 9 min read Pro lesson
- Harmful Content Detection System Design 11 min read Pro lesson
- Harmful Content Detection System Design 8 min read
- Harmful Content Detection System Design 10 min read Pro lesson
Module 6: Case Study - Video Recommendations
Case Study - Video Recommendations - 5 topics.
- 01 Problem & Discovery: Video Recommendation System 5 min read Pro lesson
- 02 ML Framing & Data 8 min read Pro lesson
- 03 Model Development 11 min read Pro lesson
- 04 Evaluation & Serving 6 min read Pro lesson
- 05 Summary & Variations 8 min read Pro lesson
Module 7: Case Study - Visual Search
Case Study - Visual Search - 5 topics.
- Visual Search System: Problem and Discovery 6 min read Pro lesson
- Visual Search: ML Framing and Data 7 min read Pro lesson
- Visual Search: Model Development 9 min read Pro lesson
- Visual Search: Evaluation and Serving 7 min read Pro lesson
- Visual Search: Summary and Variations 8 min read Pro lesson
Module 8: Case Study - Ad Click Prediction
Case Study - Ad Click Prediction - 5 topics.
- Ad Click Prediction: Problem & Discovery 6 min read Pro lesson
- Ad Click Prediction: ML Framing & Data 9 min read Pro lesson
- Ad Click Prediction: Model Development 11 min read Pro lesson
- Ad Click Prediction: Evaluation & Serving 7 min read Pro lesson
- Ad Click Prediction: Summary & Variations 9 min read Pro lesson
Module 9: Case Study - People You May Know
Case Study - People You May Know - 5 topics.
- People You May Know: Problem & Discovery 6 min read Pro lesson
- ML Framing & Data Collection 8 min read Pro lesson
- Model Development: Candidate Generation & Ranking 10 min read Pro lesson
- Evaluation & Serving 7 min read Pro lesson
- Summary & Variations 9 min read Pro lesson
Module 10: Case Study - Event Recommendations
Case Study - Event Recommendations - 5 topics.
- Event Recommendation System: Problem & Discovery 6 min read Pro lesson
- ML Framing & Data 8 min read Pro lesson
- Model Development 10 min read Pro lesson
- Evaluation & Serving 6 min read Pro lesson
- Summary & Variations 8 min read Pro lesson
Module 11: Case Study - Video Search
Case Study - Video Search - 5 topics.
- Module 11: Video Search Engine - Problem & Discovery 5 min read Pro lesson
- Module 11: Video Search Engine - ML Framing & Data 7 min read Pro lesson
- Module 11: Video Search Engine - Model Development 11 min read Pro lesson
- Module 11: Video Search Engine - Evaluation & Serving 7 min read Pro lesson
- Module 11: Video Search Engine - Summary & Variations 7 min read Pro lesson
Module 12: Case Study - Street View Blurring
Case Study - Street View Blurring - 5 topics.
- 01 Problem and Discovery: Street View Privacy Blurring 5 min read Pro lesson
- 02 ML Framing and Data 8 min read Pro lesson
- 03 Model Development 11 min read Pro lesson
- 04 Evaluation and Serving 7 min read Pro lesson
- 05 Summary and Variations 9 min read Pro lesson
Module 13: Case Study - Similar Listings
Case Study - Similar Listings - 5 topics.
- Module 13: Similar Listings on Vacation Rental Platforms 5 min read Pro lesson
- Module 13: Similar Listings on Vacation Rental Platforms 8 min read Pro lesson
- Module 13: Similar Listings on Vacation Rental Platforms 9 min read Pro lesson
- Module 13: Similar Listings on Vacation Rental Platforms 6 min read Pro lesson
- Module 13: Similar Listings on Vacation Rental Platforms 6 min read Pro lesson
Module 14: Cheat Sheets
Cheat Sheets - 3 topics.
- Interview Day Cheat Sheet 7 min read Pro lesson
- Common Patterns Reference 10 min read Pro lesson
- Mistakes to Avoid 15 min read Pro lesson