This chapter explores how Large Language Models (LLMs) can serve as end-to-end recommender systems. It covers key techniques such as prompting, fine-tuning, and cost optimization strategies including distillation, quantization, and caching. Practical design considerations are discussed to balance quality, cost, and scalability. The chapter concludes with two hands-on tutorials: one on fine-tuning LLMs for personalized movie recommendations, and another on applying knowledge distillation for efficient inference, offering practical insights for building LLM-based recommenders.