Inspiration

Study materials are scattered across textbooks, slides, and lecture recordings. We built EduMind AI to help students connect these resources, find reliable explanations, and decide what to study next.

What it does

EduMind AI turns uploaded materials into a personal learning workspace. Students can ask questions and receive source-cited answers, take adaptive quizzes, and follow personalized study plans. Video citations jump to the relevant lecture moment, while searchable transcripts and timestamped notes make revision easier. A Knowledge Galaxy displays topic practice scores.

How we built it

We used Next.js, TypeScript, and Tailwind CSS, adapting Google Stitch designs for mobile and desktop. Supabase handles authentication, private storage, and PostgreSQL with pgvector. Gemini generates embeddings, grounded explanations, and quizzes. A separate FFmpeg and Whisper pipeline processes lecture audio into timestamped transcripts.

Challenges we faced

The main challenges were preserving accurate timestamps, keeping answers grounded in sources, protecting each student's data, and handling temporary AI service failures. We added timestamp validation, ownership checks, background processing jobs, and clearly labeled fallbacks.

Accomplishments

We deployed the web app on Vercel, published the source on GitHub, and passed 15 automated tests. Integration checks also verified private video uploads, transcription, retrieval, citations, quizzes, and revision links.

What we learned

A useful learning tool needs more than generated answers. Traceable sources, focused revision, and clear feedback make the experience more valuable.

What's next

We plan to host the transcription worker separately, improve multilingual recognition, add OCR for scanned documents, and strengthen accessibility and production safeguards. The deployed app currently supports manual transcript uploads; automatic transcription requires the separate worker.

Built With

  • adapting-google-stitch-designs-for-mobile-and-desktop.-supabase-handles-authentication
  • add-ocr-for-scanned-documents
  • and-follow-personalized-study-plans.-video-citations-jump-to-the-relevant-lecture-moment
  • and-handling-temporary-ai-service-failures.-we-added-timestamp-validation
  • and-lecture-recordings.-we-built-edumind-ai-to-help-students-connect-these-resources
  • and-postgresql-with-pgvector.-gemini-generates-embeddings
  • and-quizzes.-a-separate-ffmpeg-and-whisper-pipeline-processes-lecture-audio-into-timestamped-transcripts.-##-challenges-we-faced-the-main-challenges-were-preserving-accurate-timestamps
  • and-strengthen-accessibility-and-production-safeguards.-the-deployed-app-currently-supports-manual-transcript-uploads;-automatic-transcription-requires-the-separate-worker.##-inspiration-study-materials-are-scattered-across-textbooks
  • and-tailwind-css
  • background-processing-jobs
  • citations
  • find-reliable-explanations
  • private-storage
  • protecting-each-student's-data
  • published-the-source-on-github
  • quizzes
  • retrieval
  • slides
  • take-adaptive-quizzes
  • transcription
  • typescript
Share this project:

Updates

Submission history