Inspiration
Modern learning was very boring, so I decided to fight this prolem. Reading big PDFs were annoying and took a lot of time. I wanted to save time and transform static studying into an interactive training. That's how Edugym was born - personal AI assistant for learning.
What it does
Material upload: Easily import study documents or ebooks Smart chats: Conversations with AI based strictly on the uploaded materials Test generation and taking: Automated creation of exams (simple multiple-choice or open-ended-questions) to test knowledge and analyze user's weaknesses
How we built it
Backend: Developed on FastAPI using asynchronous SQLAlchemy with pgvectors for semantic search. Frontend: Built on React, TailwindCSS, Herocoins for modern, adaptive interface. AI: Gemini API key for analyzing materials, generating exams and for giving a review after checking answers
Challenges we ran into
Interface adaptability: Designing complex cards featuring statuses and buttons for mobile devices required fine-tuning CSS grids and flexbox containers
Accomplishments that we're proud of
Smart Material Processing & Chat Integration: Implemented asynchronous database sessions with SQLAlchemy to manage dynamic study materials and power real-time, context-aware AI chat sessions based strictly on uploaded documents. Automated Exam Generation: Developed an intelligent testing engine capable of generating custom exams, handling both multiple-choice and open-ended questions, and using AI to automatically score and provide detailed feedback.
What we learned
Improved skills in REST API and FastAPI and working with async sessions in SQLAlchemy. Learned how to effectively connect frontend and backend
What's next for Edugym
Progress Analytics: Adding visual charts and statistics to track learning streaks, test scores over time, and topic mastery directly on the dashboard. Collaborative Study Spaces: Introducing features that allow students to share study materials and tests with peers while keeping personalized AI chat histories private.
Built With
- docker
- fastapi
- pgvector
- postgresql
- react
- tailwindcss
