Case Study: Design a Deep Research Agent
A full AI-engineering interview answer for a multi-step research-report agent: clarify-and-plan, parallel sandboxed sub-agents, reasoning-model allocation, synthesis with a citation agent, budget/termination controls, and evaluating the finished report
Model interview answer for designing a 'deep research' agent that produces a multi-step research report rather than a quick answer: a three-stage architecture (clarification and planning, parallel sub-agent execution in a sandbox, synthesis with redundancy removal and a dedicated citation agent), the model-tier allocation decision at each stage (cheap model vs. reasoning model), budget and termination controls for the overall multi-stage pipeline, evaluating a finished report on coverage, faithfulness and citation precision, and a worked example allocating cost across a 4-5 sub-question research query.
Practice questions (10)
-
View →
Judging a Sub-Question Decomposition
Advanced · Free -
View →
Where to Spend the Reasoning Model
Advanced -
View →
Aggregate vs. Per-Sub-Agent Budgets
Advanced -
View →
Coverage, Faithfulness, and Citation Precision Are Independent
Advanced -
View →
Reworking the Cost Allocation Table
Advanced