Inspiration
Students study from many formats: PDFs, slides and lecture videos. Switching between them makes it hard to revise and test understanding.
What it does
StudyMate AI lets students upload up to 5 study sources (PDFs, PowerPoint files, public YouTube videos). Students can ask questions about the combined material, generate 5-question MCQ quizzes, and track their progress on a learning dashboard. Quiz difficulty adapts to previous quiz results.
How we built it
Built with Python and Streamlit. PDF and PPTX text is extracted with pypdf and python-pptx. YouTube videos are passed directly to the Gemini API as video input, so videos without transcripts also work. Gemini powers the tutor chat and quiz generation.
Challenges we ran into
YouTube transcripts were not available for some lecture videos, so our first approach failed. We switched to sending public YouTube URLs directly to the Gemini API as video input, which also works for videos without transcripts.
Combining PDFs, PowerPoint files and videos under one limit of 5 sources was tricky, because each source type is handled differently.
Getting the AI to return quiz questions in a clean, parseable JSON format also needed careful prompting and error handling.
As a solo beginner, setting up Git, GitHub and API key security for the first time was a big challenge.
Accomplishments that we're proud of
- A working app that learns from PDFs, PowerPoint slides and YouTube lecture videos together.
- An AI tutor that answers from the student's own materials.
- AI-generated quizzes with adaptive difficulty and a progress dashboard.
- Built solo, as my first hackathon and first AI project, and published on GitHub.
What we learned
I learned how to use the Gemini API, including direct video understanding through the Interactions API, how to extract text from PDFs and PPTX files, how to build interactive apps with Streamlit, and how to keep API keys secret using secrets files and .gitignore. I also learned Git and GitHub basics and how to debug errors step by step.
What's next for StudyMate AI
Timestamp-based source citations, saved study history, and support for more file types.
Log in or sign up for Devpost to join the conversation.