Differentiate the core components of LLMs by examining transformers, embeddings, attention, and the training lifecycle. Develop foundational skills by exploring tokenization, visualizing attention, and comparing proprietary and open-source LLMs. Build practical capabilities by assessing model outputs for bias, safety, and misuse risks.

Analyze the anatomy of an effective prompt by examining role, task, context, format, and constraints. Develop foundational skills by constructing structured prompts and implementing zero-shot, one-shot, and few-shot prompting. Build practical capabilities by adapting prompts for common tasks, diagnosing weak prompts, and designing reusable templates.

Compare advanced reasoning strategies by examining chain-of-thought, tree-of-thought, self-consistency, generated-knowledge, and least-to-most prompting. Develop foundational skills by implementing each technique and contrasting direct and chain-of-thought answers. Build practical capabilities by combining techniques for complex tasks and justifying strategy selection.

Evaluate the risks of adversarial prompting by examining how prompt injection attacks exploit LLM applications. Develop foundational skills by simulating an injection attack and implementing defences against it. Build practical capabilities by integrating human-in-the-loop review into an end-to-end prompted application.

Evaluate the shift from prompt engineering to context engineering by examining the context window, retrieved context, and conversational memory. Develop foundational skills by structuring context, incorporating retrieved information, and maintaining memory across turns. Build practical capabilities by optimizing token budgets, compressing context, and composing context-engineered LLM solutions.