BigKunLun/Agent-Interview-100

What it solves

This project addresses the problem of learning AI Agent development in a scattered, aimless way. Instead of reading long documentation without direction, it provides 100 carefully selected questions that serve as learning anchors. Each question tells you what to learn and why, covering the entire AI Agent stack from core concepts to production practice. It also doubles as a structured interview preparation tool for AI Agent and LLM engineering roles.

How it works

The repository organizes 100 questions into 11 themed modules, arranged in a progressive four-stage path: first basic architecture (Agent fundamentals), then core abilities (RAG, tool use, multi-agent, memory, planning and reasoning), then engineering practice (prompting, evaluation, safety), and finally production ecosystem (deployment, frameworks). Each article follows a consistent four-part structure: a short answer, a detailed explanation, common mistakes and follow-up interview questions, and references. Many articles include Python code examples, pseudo-code, architecture diagrams, and citations to recent research and industry practices. The project also introduces 7 special question types, such as scenario troubleshooting, quantitative calculations, system design, code review, progressive questioning chains, and 2026 frontier topics like Computer Use, Deep Research, and EU AI Act compliance. A knowledge map and suggested learning paths help different roles (engineers, architects, security specialists) navigate the content efficiently.

Who it's for

This is for anyone who wants to systematically learn AI Agent technology: developers seeking a solid foundation, architects designing Agent systems, researchers wanting an engineering perspective, job candidates preparing for AI Agent or LLM interviews, and technical interviewers looking for a ready-made question bank. The difficulty levels (basic, intermediate, advanced) and role-specific paths make it adaptable to varied backgrounds.

Highlights

• 100 curated questions covering the whole AI Agent stack, from basics to production
• 11 themed modules with a clear knowledge map and progressive learning order
• Four-part article structure (short answer, details, pitfalls/follow-ups, references) for efficient study
• Rich code examples with Python, pseudo-code, and architecture diagrams
• Practical special question types like scenario debugging, quantitative metrics, system design, and code review
• Up-to-date content referencing 2024-2026 research and industry trends, including Computer Use, Deep Research, A2A protocol, and EU AI Act compliance