Lumen AI Tutor

Your adaptive AI tutor that explains any concept at the level you pick with diagrams, quizzes, and flashcards.

Comment

Inspiration

Every learner understands differently. Standard textbooks often explain complex topics with a one-size-fits-all approach. Lumen was built to solve this by providing an adaptive AI tutor that tailors explanations, diagrams, and quizzes precisely to the user's preferred depth.

What it does

Lumen dynamically adapts its teaching strategy based on three distinct depth modes:

  • Kid Mode: Uses friendly analogies and simplified breakdowns.
  • Student Mode: Delivers structured, textbook-style clear explanations.
  • Expert Mode: Provides deep technical mechanics, rigorous equations, and microarchitectural details.

Key Features:

  • Instant Diagram Generation: Automatically creates labeled Mermaid diagrams to visualize concepts.
  • Interactive Quizzes: Tests understanding with dynamically generated multiple-choice questions and instant detailed feedback.
  • Quick-Review Flashcards: Summarizes key concepts into expandable flashcards for revision.
  • Multimodal Homework Analysis: Accepts image uploads of textbook pages, notes, or problems to analyze and explain.
  • Persistent Sessions: Simple custom access codes let users save and resume learning sessions anytime across devices.

How we built it

  • Frontend UI: Built with Streamlit for a fast, clean mobile-first web interface.
  • LLM Engine: Powered by Groq API running llama-3.3-70b-versatile for blazing-fast response generation.
  • Database & Persistence: Firebase Realtime Database for managing cross-device session states.
  • Diagram Rendering: Mermaid.ink API for live visual flowchart rendering.

Challenges we ran into

Balancing ultra-fast inference speed while generating strict structured JSON for quizzes and Mermaid syntax for diagrams without breaking the rendering pipeline.

Accomplishments that we're proud of

Successfully building a seamless, responsive mobile UI that can generate diagrams, quizzes, flashcards, and handle vision inputs—all with instant responses using Groq.

What we learned

Optimizing prompt engineering techniques for strict JSON structure output and mastering multi-level adaptive explanation models.

What's next for Lumen

Expanding multimodal capabilities to support audio explanations, flashcard export features, and personalized learning path analytics over time.

Built With

Share this project:

Updates

Submission history