My project, JEDI, is a personalized learning assistant designed to help learners efficiently navigate complex topics by providing structured learning paths, curated resources, video recommendations, quizzes, and AI-powered explanations. The inspiration for this project came from noticing how overwhelming it can be for students and self-learners to figure out where to start and what to study next, especially for technical subjects or niche fields. I wanted to create a tool that simplifies learning and makes it more interactive and goal-oriented.
Through building this project, I learned a lot about integrating multiple technologies: React for a responsive frontend, FastAPI for a fast backend, a llama3 LLM model for summarization and chat features, and web scraping techniques to pull content from YouTube and online articles. One key learning was how to design a system that connects these different components smoothly while keeping the interface intuitive for the user.
I built the project by first structuring the learning path backend, generating subtopics, and linking relevant resources and videos. On the frontend, I implemented collapsible phases, progress tracking, quizzes, and a chat feature where users can ask questions about a topic. A highlight of the project is the AI-powered summarization feature that can take a webpage or article link and produce concise, digestible points for the learner.
The main challenges I faced included handling asynchronous data fetching efficiently, ensuring smooth communication between the frontend and backend, and dealing with variability in external content (like YouTube videos or webpage formats). Additionally, fine-tuning the summarization output to be clear and useful required careful prompt engineering and testing. Overcoming these challenges taught us a lot about debugging, API integration, and user experience design.
Overall, JEDI is the result of combining AI, web development, and educational design into a single tool aimed at making learning more structured, interactive, and effective.
Built With
- css
- fastapi
- fetchapi
- html
- javascript
- json
- llama3
- ollama
- python
- tailwind
Log in or sign up for Devpost to join the conversation.