prompt engineering
Designing, testing and refining prompts so a model produces the output you need reliably.
In more detail
It covers structure, examples, reasoning instructions and output formats, and it only counts as engineering when each change is measured against a labelled eval set. Context engineering widens the same work to everything the model reads.
Related terms
- prompt — The text and other content you send a model to get a reply: instructions, context, examples and the request itself.
- context engineering — Choosing and arranging everything a model reads on each call, from instructions and tools to history, memory and retrieved data, within a token budget.
- AI eval — A repeatable test of an AI feature: a set of inputs, a way to grade each output, and a score you track across changes.
Taught in
- Prompts As Code — AI Engineering: LLM Apps, RAG & Agents
- From Prompts To Context — AI Engineering: LLM Apps, RAG & Agents