Quillium

Inspiration

Education today still relies heavily on static learning materials—long PDFs, dense notes, and passive reading. For many students, especially during short study windows or last-minute revision, this approach simply doesn’t work.

These challenges are amplified when:

• Content is not in a student’s native language
  
• Students don’t know how to test their understanding effectively
  
• Learners lack real-time, personalized feedback
  

Steel City Hacks focuses on using technology and AI to improve education, making learning more personalized, inclusive, and engaging. Quillium was inspired by this exact goal.

We wanted to explore how modern AI models like Google Gemini can transform passive study materials into active, personalized learning experiences—quickly and accessibly.

At its core, Quillium is built on a simple belief:
Education should adapt to learners, not force learners to adapt to rigid systems.

---

What it does

Quillium is an AI-powered learning assistant that converts static PDF documents into interactive study tools.

With Quillium, students can:

• Upload PDFs such as notes, textbooks, or slides
  
• Automatically generate meaningful MCQs using AI
  
• Study with AI-generated flashcards
  
• Ask questions through an AI tutor (“Ask Quill”)
  
• Track their learning progress
  
• Generate short-form study scripts and optional audio narration (“Study Shorts”)
  

Instead of passively rereading content, learners actively engage with material in formats designed to improve understanding, retention, and accessibility.

---

How we built it

Backend

• FastAPI (Python) for a clean, scalable backend
  
• PyMuPDF for extracting structured text from PDFs
  
• Google Gemini used for:
  • MCQ generation
    
  • Flashcard creation
    
  • Short-form explanations
    
  • AI tutoring responses
    
• OpenAI Text-to-Speech for optional audio narration
  
• Modular REST APIs for each learning feature
  

Frontend

• Next.js (App Router) with React and TypeScript
  
• Tailwind CSS for a clean, modern, student-friendly UI
  
• Dedicated sections for:
  • Upload
    
  • Quizzes
    
  • Flashcards
    
  • Progress tracking
    
  • Study Shorts
    
• One-page navigation for fast interaction
  
• LocalStorage-based progress tracking
  

AI Flow

PDF upload → text extraction → Gemini prompts → MCQs → flashcards → study shorts → progress tracking

---

Challenges we ran into

• Extracting clean, structured text from varied PDF formats
  
• Ensuring AI-generated questions were meaningful and not repetitive
  
• Keeping explanations concise and student-friendly
  
• Balancing advanced AI capabilities with an intuitive UI
  
• Designing learning outputs that work across different languages
  

We addressed these challenges by refining prompts, adding validation layers, and testing with multiple documents and languages.

---

Accomplishments that we're proud of

• Built a complete end-to-end AI-powered learning prototype within the hackathon timeframe
  
• Successfully integrated Google Gemini across multiple educational features
  
• Transformed PDFs into quizzes, flashcards, and short learning scripts
  
• Added multilingual learning support (50+ languages), improving inclusivity
  
• Designed a clean and engaging interface that promotes active learning
  

Most importantly, we built a tool that feels genuinely useful for students, not just technically impressive.

---

What we learned

• AI is most effective when it supports thinking, not replaces it
  
• Prompt design directly impacts educational quality
  
• Short-form, interactive learning improves engagement and retention
  
• Accessibility and UX are as important as model performance
  
• Rapid experimentation is essential when building AI-first educational tools
  

---

What's next for Quillium

• Adaptive quizzes that adjust difficulty based on student performance
  
• Captions and subtitles for Study Shorts
  
• Audio-only learning modes for accessibility
  
• Collaborative study rooms for peer learning
  
• Scalable deployment for classrooms, educators, and institutions
  

Quillium began as a hackathon project, but it has strong potential to grow into a practical, inclusive, AI-powered learning companion aligned with the future of education.

Built With

• fastapi
• googlegemini
• next.js
• pymupdf
• python
• react
• tailwindcss
• typescript