What is AI Engineering?
AI Engineering combines AI knowledge, software engineering, effective use of coding agents and the judgement required to decide what should be built.
The field is not only model prompting. It includes shaping the user problem, building model-backed application behaviour, engineering reliable software around probabilistic components, and using coding agents without giving up human responsibility for quality and risk.
The organising frame for this knowledge base is Andrew Ng’s AI Engineering Skills Map, published in DeepLearning.AI’s The Batch. This site builds from Ng’s four-area framing while making explicit where the curriculum extends the source material.
Another useful lens is the distinction between Software Engineering for AI and AI for Software Engineering. One direction asks how to engineer systems that contain AI. The other asks how to use AI to perform software engineering. Modern AI engineering increasingly combines both.
This view frames AI engineering as a production loop: teams shape the problem, design the behaviour, build the model-backed artifact, evaluate it, deploy it, operate it and learn from evidence. The artifact includes model, data, prompts, tools, evals and guardrails because production quality depends on the whole system.
This view treats AI engineering as architecture around probabilistic capability. The model is only one component. User experience, application logic, model gateways, retrieval, tools, evaluation, observability and governance create the control points that make behaviour useful, measurable, secure and changeable after deployment.