Personal project · Case study
AI Wiki Quiz Generator
Paste a Wikipedia link, get a quiz.
An AI-powered tool that generates multiple-choice questions from any Wikipedia article using Google Gemini and LangChain, then stores every quiz so it can be revisited.
- Role
- Personal project (backend, AI pipeline, frontend)
- Stack
- FastAPI
- React
- PostgreSQL
- LangChain
- Google Gemini
- BeautifulSoup
- SQLAlchemy
- Tailwind CSS
01 · Context
The problem
Turning an article into a good set of quiz questions by hand is slow, and a raw LLM call alone does not give you clean source text or a place to keep the results.
02 · Solution
The approach
A three-step pipeline: scrape and clean the article text, have Gemini (orchestrated with LangChain) write the questions, then persist the quiz so the history and individual quizzes can be reopened later.
03 · Architecture
How the pieces fit together
AI Wiki Quiz Generator · architecture
Interface
- React UI · quiz · history · dark mode
Logic
- FastAPI · generate · history · quiz by id
- Scraper · BeautifulSoup
Data
- PostgreSQL · quizzes · history
Services
- LangChain → Gemini · question generation
Data flow
- React UI → FastAPI (URL)
- FastAPI → Scraper (fetch)
- FastAPI → LangChain → Gemini (prompt)
- FastAPI → PostgreSQL (save)
04 · What I built
Key features
01
Scrape → generate → store
BeautifulSoup extracts clean text, LangChain and Gemini produce the questions, and PostgreSQL persists them.
02
REST API
FastAPI endpoints to generate a quiz, list history and fetch a quiz by id.
03
Interactive quiz mode
A React UI where you can take a quiz and get instant scoring, with dark mode.
04
History
Previously generated quizzes are saved and can be reviewed at any time.