Prompting vs RAG vs Tuning Under Real Constraints

Prompting vs RAG vs Tuning Under Real Constraints

Path Outline

Progress0%

1The FDE Design-Round Rubric Under Time Pressure 2Five-Minute Discovery to a Scoped Problem Statement 3Building an Estimation Sheet From Load to Quota Risk 4Narrating the End-to-End Request Path With Failure Recovery

5Solution Class and Build-vs-Buy Boundaries

6Prompting vs RAG vs Tuning Under Real Constraints

7Predictive ML and LLM Decision Boundaries 8Case Integration — Northwind Architecture v1

9Document Intake and Index Freshness Design 10Permission-Aware Retrieval Under a Context Budget 11Case Integration — Northwind Knowledge Assistant

12Single-Agent Execution With ADK State 13Multi-Agent Patterns and Failure Modes 14Tool Boundaries With MCP and A2A 15Deterministic Human Approval Checkpoints 16Case Integration — Northwind Claims Agent

17Identity Propagation and Trust Boundaries 18Regulatory Obligations as Design Controls 19Northwind Security Layers and Abuse Defenses

20End-to-End Latency Budgeting for Agents 21Fallback-Mode Tables and Model-Tier Routing 22Evaluation Pipelines From Offline to Online 23Diagnosis Drill — Northwind Quality Drop

24Mock Interview: Retail Analytics on BigQuery 25Mock Interview : Telecom Voice-Plus-Chat Service 26Mock Interview : Ambiguous Prompt Under Pressure

Hide outline

Home
Paths
Google Cloud FDE System Design in Depth
Prompting vs RAG vs Tuning Under Real Constraints