Slide Analyzer Project
An open source AI-based slide validator designed for conference organizers. It validates submitted slides in three steps: first by checking the format (PDF, Google Slides, Keynote, Figma, Canva, or URL), then running deterministic checks (deck size, slide count, fonts, media presence), and finally executing AI-powered probabilistic assessments (e.g., verifying a title slide, checking for appropriate content layout). The project features a responsive user interface that provides real-time feedback with loading indicators for each test, and an admin panel for configuring accepted file types and validation criteria. Once all tests pass, valid decks are integrated into a master deck.
Features
- Multi-format slide validation (PDF, Google Slides, Keynote, Figma, Canva, URL)
- Deterministic technical checks for deck specifications
- AI-powered content assessment
- Real-time feedback with progress indicators
- Admin configuration panel
- Master deck integration
Technical Stack
- Python
- FastAPI
- Uvicorn
- Next.js
- React
- Tailwind CSS
- TypeScript
Lessons Learned
Clean separation between backend and frontend enhances maintainability. Modern tech stack and custom deployment configurations (Replit, repomix) streamline development and deployment. UI components like slide previews and progress bars provide a user-friendly experience.