AI engineering
Building software features on models someone else trained: prompts, retrieval, tools and agents, with evals, monitoring and safety around them.
In more detail
An AI engineer treats the model as a component that varies, can be wrong fluently, costs money per call and can change underneath you. Most of the work is ordinary software engineering around that component: deciding whether a model is needed at all, measuring its output, and limiting what it can do.
Related terms
- large language model — A neural network trained on huge amounts of text that generates text by predicting one token at a time from everything before it.
- 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.
- retrieval-augmented generation — Answering a question by first retrieving relevant passages from your own documents and adding them to the prompt, so the model answers from them.
Taught in
- What An AI Engineer Does — AI Engineering: LLM Apps, RAG & Agents