Recommender Systems
Metrics and offline evaluation, collaborative filtering, candidate generation, ranking and recommendations in production
Open "Fundamentals & Metrics"
Fundamentals & Metrics
Problem framing, implicit feedback, ranking metrics, offline evaluation and why it diverges from online
10 questions
Open "Collaborative Filtering"
Collaborative Filtering
User- and item-based neighbours, matrix factorization, ALS and BPR, hybrids with content
10 questions
Open "Candidate Generation & Embeddings"
Candidate Generation & Embeddings
Two-tower models, approximate nearest neighbours, multiple candidate sources and blending
10 questions
Open "Ranking & Re-ranking"
Ranking & Re-ranking
Features, CTR prediction and calibration, multi-objective ranking, diversity and business rules
10 questions
Open "Production: Bias, Cold Start, Experiments"
Production: Bias, Cold Start, Experiments
Feedback loops, position bias, cold start, A/B tests and bandits, monitoring
10 questions