Documentation¶
Last reviewed: October 2026
Practical, cross-cutting reference pages for building and operating GenAI systems in production: architecture, retrieval, agents, model choice, prompting, security, troubleshooting and quick answers. They complement the per-topic deep dives under GenAI Topics and the interview Q&A banks.
Each page follows the same shape, so you can use it both as a working reference and as interview preparation:
- Mechanics first: diagrams, decision tables and worked examples.
- Failure modes: what breaks in production and how you would notice.
- How interviewers probe this: senior-level questions with what a strong answer covers.
- Further reading: official docs, specs and primary papers only.
-
Architecture Overview
The layers of a production GenAI application, request lifecycle, and where latency, cost and risk live.
-
RAG Flow
Indexing and query pipelines step by step: parsing, chunking, hybrid search, re-ranking, context assembly and RAG evaluation.
-
Agent Workflow
The plan, act, observe loop; tool design; bounding cost and risk; single vs multi-agent; durable execution.
-
Model Selection Guide
Choosing between frontier, mid-tier, fast and open-weight models in the October 2026 landscape, plus routing and eval-driven selection.
-
Prompt Engineering Best Practices
Prompt structure, few-shot, structured output, reasoning effort, prompt caching and treating prompts as versioned code.
-
Security & Governance
Prompt injection, data leakage, agent permissions, guardrails, audit, and mapping controls to OWASP, NIST and the EU AI Act.
-
Troubleshooting
Symptom, cause and fix tables for RAG, agents, latency and cost, evals, and deployment.
-
FAQ
Short, opinionated answers to the questions that come up most in design reviews and interviews.
Suggested reading paths¶
| If you are preparing for... | Read in this order |
|---|---|
| AI / GenAI engineer loops | Architecture → RAG Flow → Prompting → Troubleshooting |
| Agent-heavy or applied AI roles | Agent Workflow → Security & Governance → Model Selection |
| Staff / principal architecture rounds | Architecture → Model Selection → Security & Governance → FAQ |
| Data platform roles moving into AI | Architecture (data stack section) → RAG Flow → Security & Governance |
Pair these pages with practice
After each page, answer its "How interviewers probe this" questions out loud before reading the strong-answer notes. Then drill the matching question bank, for example GenAI Interview Q&A, Agents Interview Q&A or Production Incidents.