learn.ai

Interview questions

Answer each one out loud, then rate yourself.

Week 1: Foundations and an agent loop from scratch

Walk through an agent loop step by step.
When is a fixed workflow better than an agent?

Week 2: RAG and context engineering

What chunking strategies do you know, and how do you pick one?
Why does RAG fail?
When is hybrid search better than pure vector search?

Week 3: MCP, LangGraph, memory and multi-agent

MCP or plain function calling? When does each make sense?
When is a multi-agent design worth its cost?
What types of memory does an agent have, and where does each live?

Week 4: Evals, security and observability

How do you know it works?
How do you defend against prompt injection?
How would you halve latency and cost?

Week 5: Ship it and prepare for interviews

Explain your project's architecture and its biggest trade-off in 3 minutes.
What breaks first at 100x traffic?