Architecture Synthesis: Choosing a System Design

Practise and Check

Deliverable

Architecture review: write a two-page design document for your capstone that justifies each component against a simpler alternative.

Assessment

Answer the five Interview questions without notes, then run the lab.

Interview probe

Larkspur's support assistant moved from plain RAG to Self-RAG. Latency is several times higher, and “What happens if my server goes down?” still gets a wrong answer. What happened, and what do you change?

Common mistakes

• Comparing designs by query-time calls while ignoring index-time LLM work
• Judging retrieval and generation from one end-to-end correctness score
• Adding Self-RAG to recover evidence that retrieval missed from the top k