This chapter explores prompt engineering as a foundational technique for leveraging large language models in diverse language service applications. It details how carefully crafted prompts define tasks, activate model capabilities, and constrain outputs, emphasizing clarity, structure, context provision, task decomposition, and example guidance. The chapter introduces advanced prompt types such as zero-shot, few-shot, Chain-of-Thought, and Tree-of-Thought, and discusses self-consistency strategies to optimize reasoning. Structured and programming-style prompts are presented as scalable solutions for complex scenarios, with modular template design supporting adaptability and automation. Through illustrative examples, the chapter demonstrates how prompt engineering bridges user intent and model performance, enabling reliable, controllable, and creative language services.