Falk-HI1107×1388 297 KB

Kim Falk

Building a recommender means making decisions across an entire pipeline: which candidates to retrieve, what to filter out, how to score the remaining options, and what to put in front of the user. Adding an LLM introduces more possibilities, but it also raises practical questions about where it belongs and how to evaluate its contribution.

That’s the territory this book explores. It walks through candidate generation, filtering, scoring, and reranking, examining where LLMs can help. It also covers how to use content metadata and user behavior, integrate custom prompts, RAG, and agents, and apply these approaches to both new designs and existing systems.

The focus extends beyond getting a promising prototype running. You’ll learn about evaluating recommendations against the use case they’re meant to support, handling cold-start problems, and deploying a customer-facing system with the monitoring it needs.

Kim is also the author of Practical Recommender Systems and has spent a decade and a half working on recommenders across news, media, and e-commerce. His perspective is grounded in making research work in production.

The book is aimed at developers, data workers, and researchers with basic Python, statistics, and machine learning knowledge. If you’re building a recommender or considering how newer AI approaches might fit into one you already maintain, take a look.

• Full details: AI Recommender Systems - Kim Falk

Don’t forget you can get 45% off with your Devtalk discount! Just use the coupon code “devtalk.com” at checkout