Book series
Recommendation Systems
MOST RECOMMENDATION SYSTEMS FAIL BEFORE THEY SERVE A SINGLE PREDICTION The problem is rarely the algorithm—it's the chasm between academic architectures and production realities....
3 books
Recommendation Systems
MOST RECOMMENDATION SYSTEMS FAIL BEFORE THEY SERVE A SINGLE PREDICTION The problem is rarely the algorithm—it's the chasm between academic architectures and production realities. That's why this series exists: to bridge that gap across the entire stack.
📘 Book 1: Recommendation Systems Foundations Lays the ground truth: recommendation is not a single algorithm but an integrated pipeline. Covers data foundations, classical methods like matrix factorization, multi-stage architectures, and modern deep learning—all with engineering trade-offs and production realities in mind.
You'll learn to think in terms of latency, scale, business alignment, and evaluation. 📗 Book 2: Engineering Recommendation Models Dives deep into the modeling layer: retrieval, ranking, sequential recommendation, multi-objective optimization.
Framework-agnostic, it dissects architectures like two-tower, Wide & Deep, DIN, SASRec, and MMoE. You'll diagnose production failures like popularity bias and feedback loops.
📙 Book 3: Building Large-Scale Recommendation Systems Takes you from offline training to online serving at scale. Covers distributed caching, A/B testing, continuous learning, and case studies from YouTube, Netflix, Amazon, and more.
Focuses on latency, throughput, cost, and reliability—the engineering constraints that make or break a system. This series is for engineers who want to move beyond theory.
Each book is packed with system diagrams, trade-off...
Books in this series
Book 1
Recommendation Systems Foundations: Principles, Algorithms, and Modern Recommendation Pipelines
Ryan Mercer
Book 1
Recommendation Systems Foundations: Principles, Algorithms, and Modern Recommendation Pipelines
Ryan Mercer
Many developers treat recommendation as a single algorithm problem—train a model, serve predictions. In reality, production recommenders are integrated pipelines where data quality, retrieval speed, ranking precision, and business goals must coexist. This boo...
Book 2
Engineering Recommendation Models: Deep Learning, Ranking, and Advanced Recommendation Techniques
Ryan Mercer
Book 2
Engineering Recommendation Models: Deep Learning, Ranking, and Advanced Recommendation Techniques
Ryan Mercer
Most recommendation systems fail before they even serve a single prediction. The problem is rarely the algorithm—it's the chasm between academic architectures and production realities. Engineering Recommendation Models: Deep Learning, Ranking, and Advanced R...
Book 3
Building Large-Scale Recommendation Systems: Production Architecture, Scalability, and Real-World Systems
Ryan Mercer
Book 3
Building Large-Scale Recommendation Systems: Production Architecture, Scalability, and Real-World Systems
Ryan Mercer
Most machine learning engineers can train a recommendation model that achieves impressive offline metrics, but getting that same model to serve personalized recommendations to millions of users within milliseconds — without crashing, degrading, or burning thr...