Retrieval-Augmented Generation (RAG) & Knowledge Systems for Government Agencies (Advanced) (Self-Paced)
IDTECH9190SP

Module 1: Why RAG, and how it works

• The grounding problem
• RAG architecture end to end
• Government use cases: policy, records, knowledge bases
• What RAG does and doesn't fix

Module 2: Preparing your content

• Chunking strategies
• Cleaning and structuring documents
• Metadata and source tracking
• Handling sensitive content

Module 3: Embeddings and retrieval

• Embeddings and vector stores
• Similarity search
• Tuning retrieval quality
• Keeping sources current

Module 4: Grounded answers and evaluation

• Assembling answers with citations
• Reducing hallucination
• Evaluating accuracy
• Handling 'I don't know'

Module 5: Plan a knowledge system (capstone)

• Scoping a RAG use case
• Designing the pipeline
• An evaluation plan
• A path to a pilot