Zchary1106/agent-interview-hub

What is this?

Agent Interview Hub is a free, open-source study library for engineers preparing for AI Agent job interviews at major Chinese tech companies (Alibaba, ByteDance, Tencent, Baidu, Meituan, Huawei, etc.) and top overseas AI firms (OpenAI, Anthropic, Google DeepMind, Microsoft).

It's essentially a giant, organized collection of interview prep material — questions, answers, company-specific guides, and real interview reports — all in one place.

What's inside

• 300+ interview questions with reference answers — covering everything from core concepts to advanced system design
• 15 categories of company/position materials — each major company has its own folder with job requirements, interview questions, and (where available) real interview experiences shared by candidates
• 6 timed practical coding challenges — realistic take-home style tasks (e.g., building a document Q&A agent, a multi-agent collaboration system, a ReAct-pattern agent) with difficulty ratings and time limits (1–6 hours)
• A 16-week learning roadmap — a week-by-week study plan from basics to job offer
• 35 general knowledge documents — deep-dive write-ups on core topics

What topics does it cover?

The knowledge is organized into four layers:

1. Basics — Transformer architecture, LLM fundamentals, prompt engineering
2. Core — Agent design patterns (ReAct, Plan-and-Execute, Multi-Agent), RAG pipelines, function calling, MCP (Model Context Protocol), frameworks like LangChain and LangGraph
3. Advanced — Agentic RAG, GraphRAG, model fine-tuning, inference optimization, agent security and evaluation
4. Practical — system design, company job requirement breakdowns, interview techniques, and real interview reports

How is it different from a typical question bank?

• Every question has an answer — not just a list of prompts
• Real interview reports — sourced from public posts on Niuke (牛客), Xiaohongshu, Zhihu, CSDN, GitHub, and similar platforms
• Company-specific folders — you can study exactly what a given company tends to ask
• A built-in "Interview Collector Agent" — a script that can automatically search public sources for new interview reports and organize them into structured candidates, installable into tools like GitHub Copilot CLI, Claude Code, or Cursor

How do you use it?

Three ways:

1. Read online — the project is published as a static website on GitHub Pages (mobile-friendly), with a classic version, a newer "exam-prep path" version, and a "high-frequency question signal board"
2. Clone locally — git clone and open the Markdown files in any editor
3. Build your own static site — run a Python script that generates a dist/ folder, then serve it locally

License

MIT — free to use, modify, and share.

Bottom line

If you're a Chinese engineer interviewing for AI Agent roles at big tech companies, this is a one-stop, community-maintained prep resource: structured learning path, real questions with answers, company-specific intel, and practical coding challenges — all free and open source.