This chapter covers key LLM techniques that address traditional recommendation challenges to enhance existing systems. It begins with an overview of the importance of using LLMs to enhance recommendation systems, then covers key techniques for LLM enhancement, including tokenization, embeddings for richer data representation, and ANN algorithms for efficient retrieval. We’ll also explore how LLMs assist in data labeling and evaluation, enabling more accurate and scalable recommendation systems. We then close the chapter with two tutorial examples, one demonstrates the use of LLMs for topic classification and item similarity labeling, and another shows how to combine LLM embeddings with traditional ranking models for news recommendation.