Quillium

Start by uploading any text-based PDF notes, textbook chapters, worksheets, anything.
Exam-style questions generated adapting to content difficulty.Instant feedback supports confidence and retention

✨ Inspiration

Students don’t struggle because they lack intelligence — they struggle because studying is still built around passive reading, not how memory actually works.

We kept seeing the same pattern everywhere:

📄 Endless scrolling 🔁 Rereading without retaining ❓ Still unsure of what they actually understood

And another problem became clear: access isn’t equal.

🌍 Some students learn in a language that isn’t their first 🧠 Some need structure, not long paragraphs ♿ Some require accessibility features like dyslexia-friendly formatting 🔥 Many just needed studying to feel less overwhelming and more adaptive

So we asked:

What if studying adjusted to the learner — not the learner to the material?

That question became Quillium.

Quillium is built around one core belief:

Learning should follow how memory works — through active recall, spaced repetition, multimodal reinforcement, and accessible design.

Not just another flashcard tool. Not another reading assistant. But a personal learning companion that turns information into understanding.

🚀 What It Does

Quillium transforms any text-based PDF into an interactive, adaptive learning system.

Once uploaded, Quillium:

✔️ Generates high-quality practice questions using evidence-based learning science

✔️ Creates adaptive flashcards that adjust to learner performance

✔️ Translates content and questions into 50+ languages

✔️ Automatically changes difficulty as understanding improves

✔️ Provides a dyslexia-friendly reading mode

✔️ Tracks progress through simple, meaningful analytics — not confusing dashboards

Learners move through a purposeful cycle:

Understand → Practice → Recall → Improve → Master

Quillium doesn’t force learners to change how they study — it reshapes material to match how learning works.

🧠 How We Built It

Quillium blends NLP, multilingual AI, and accessibility-first UX.

Tech Stack 🛠️

📝 PyMuPDF → Extract structured text from PDFs
🤖 DeepSeek (via OpenRouter) → Generate structured MCQs
🧩 NLTK + WordNet → Smart distractors
🌍 Helsinki MarianMT → Translation across 50+ languages
🎨 Streamlit + Plotly → Accessible and elegant learner experience

⚠️ Challenges We Ran Into

🗣️ Ensuring translation accuracy across diverse languages
🎯 Generating high-quality, exam-style questions consistently
🏗️ Adding accessibility features without overwhelming the UI
🚀 Keeping performance smooth for low-spec devices and slow networks

🏆 Accomplishments We’re Proud Of

🎓 A working prototype that generates quizzes, flashcards, and translations automatically
🧬 A workflow rooted in modern memory science
🌍 Accessibility and multilingual support built from day one — not as a future feature
💬 Early testers said studying felt clearer, structured, and less overwhelming

📚 What We Learned

🌐 Multilingual AI meaningfully reduces educational barriers
🔁 Adaptive repetition increases retention and confidence
🧩 Students don’t need more content — they need learnable content
🧼 Simplicity invites use — complexity scares learners away

🔮 What’s Next

To continue building Quillium into a global learning platform, we’re working on:

📶 Full offline access for low-bandwidth regions
✍️ OCR to support handwritten and printed notes
🎤 Voice interaction + natural text-to-speech
🔗 Export to Notion, Anki, Google Forms, and LMS platforms
🧠 A fully adaptive learning engine that adjusts pacing and difficulty automatically

👉 The long-term vision: