AI Engineering Knowledge Base

Build AI systems that are useful, reliable and responsible

AI Engineering is the discipline of designing, building, evaluating, deploying and improving software systems in which AI models perform part of the system’s work.

Start with the foundationsExplore coding agents

Cycle

Four Areas Reinforce Each Other

Shape the problem, build the AI capability, engineer the system, and use coding agents to accelerate verified change.

Area 1

Build AI applications

Add model capability, context, grounding, tools, evals and production controls.

ExploreArea 2

Engineer reliable systems

Use architecture, testing, security, data and operations to make software sustainable.

ExploreArea 3

Use coding agents

Direct agents through bounded tasks, context, tools, permissions and evidence.

ExploreArea 4

Shape the build

Turn ambiguous needs into valuable, feasible and responsible interventions.

Explore

1. Shape
2. Build
3. Engineer
4. Agent-assist
5. Evaluate outcomes

Evals connect everything

Evals turn intent into evidence.

Shaping defines desirable behaviour, AI application engineering converts it into datasets and graders, software engineering automates the checks, and coding agents use the same evidence to improve work.

Learning path

Foundation path

Prior knowledge: General programming familiarity.

Outcome: Understand the four-area map and the baseline concepts needed for AI Engineering study.

1. What is AI Engineering?
2. The four-area skills map
3. Machine learning
4. Requirements and design
5. AI suitability

Learning path

Builder path

Prior knowledge: Software engineering experience and basic AI literacy.

Outcome: Design, evaluate and operate grounded AI application workflows.

1. Foundation models and LLMs
2. Context engineering
3. Grounding and RAG
4. Tool calling and MCP
5. Evals and error analysis

Learning path

Agentic development path

Prior knowledge: Comfort with repositories, tests and code review.

Outcome: Use coding agents deliberately, safely and with evidence-based completion.

1. Coding assistants versus coding agents
2. The agent loop
3. Configuration and permissions
4. The ten-step working method
5. Multi-agent orchestration

The four top-level areas follow Andrew Ng’s AI Engineering Skills Map. Ng has elaborated Area 1 in more detail; the detailed structures for Areas 2-4 are a reasoned extension for this knowledge base, not a claim of direct attribution or endorsement.