AI ENGINEER FOUNDATIONS · STAGE 05
Generative-AI application development
Build useful applications around language models.
Move from a model call to a dependable application. Learn prompt design, retrieval, document preparation, evaluation, and ways to reduce unsupported answers.
BEFORE YOU START
Prerequisites
- LLM foundations
- Basic API and data concepts
LEARNING OBJECTIVES
What you should be able to do
- Explain retrieval-augmented generation
- Design a document ingestion and retrieval flow
- Evaluate evidence and answer quality
WHAT YOU'LL COVER
Key concepts
SEE THE CONNECTIONS
A visual learning sequence
These are connected learning steps, not the only valid order. Revisit earlier ideas when an exercise reveals a gap.
Suggested learning sequence
Prepare knowledge
Extract, chunk, and index approved source documents.
Why it matters: Relevant context must be findable at query time.
Retrieve and answer
Find passages, construct context, and generate a cited response.
Why it matters: Separating retrieval from generation makes the data path explicit.
Evaluate the whole path
Measure retrieval, grounding, safety, and user outcomes.
Why it matters: A fluent response can still be unsupported.
Put it into practice
Prototype a policy assistant that retrieves the current handbook section and declines to answer when no relevant source is found.
Practice tasks
- Diagram indexing and query-time stages.
- Write a small evaluation set with expected source passages.
- Compare RAG and fine-tuning for changing facts.
How you'll know you're ready
- Can identify retrieval failure separately from generation failure.
- Can explain why access control must apply to retrieved passages.