Case studyAI / ML
AI Exam Analyzer
Analyze past papers and generate practice papers.
- Year
- 2026
- Focus
- Exam paper analysis
Overview
An exam-paper analysis pipeline for extracting questions, studying frequency and syllabus coverage, and generating mock papers.
The challenge
Past papers contain repeated questions and scanned content that need extraction and grouping before topic coverage can be analyzed.
How it works
Extract PDF text with PyMuPDF, use Tesseract OCR for scans, and parse questions with Gemini.
Use embeddings for semantic deduplication, then calculate frequency, recency, and syllabus coverage.
Generate mock papers with pattern rules and optional Gemini phrasing, exposed through FastAPI and a Next.js frontend.
Technology stack
- Python
- FastAPI
- Gemini
- PyMuPDF
- Tesseract
- Next.js
- TypeScript
Highlights
- scanned paper extraction
- OCR
- syllabus topic analysis
- Coverage
- pattern-based generation
- Mock papers