Nexus EduAI - FUTMINNA Hack Day

Nexus EduAI - FUTMINNA Hack Day .it is an intelligent campus companion. upload messy notes and take instantly clarity. Blessings

Comment

Nexus EduAI: Your Intelligent Campus Companion 🎓🤖

Built for Gemini Hack Day FUTMINNA

🎥 Video Demo Link

Insert Video Demo Link Here (Replace with your actual demo link)

💡 What inspired us?

As university students at FUTMINNA, we constantly face the overwhelming challenge of digesting massive amounts of lecture notes, deciphering messy whiteboard scribbles, and preparing for exams under tight deadlines. Traditional studying is often passive and time-consuming.

When the Gemini Hack Day FUTMINNA was announced, we saw the perfect opportunity to solve a problem that matters deeply to our campus community. We wanted to build something that doesn't just give answers, but actually helps students learn. Thus, Nexus EduAI was born—an intelligent learning companion that transforms raw, unstructured lecture materials into interactive, bite-sized study aids and provides a patient, 24/7 AI tutor.

⚙️ How we built our project

Nexus EduAI is a modern web application built with a focus on speed, user experience, and cutting-edge AI.

Tech Stack:

  • Frontend: React, TypeScript, and Tailwind CSS for a highly responsive, polished, and accessible user interface.
  • AI Engine: Google's new @google/genai SDK.

Key Features & Implementation:

  1. Smart Note Analyzer: We utilized the gemini-3-flash-preview model's multimodal capabilities. Students can upload images of their notes or paste text. We used Gemini's powerful responseSchema to force the model to return a strictly typed JSON object containing an executive summary, key entities, and an array of Q&A pairs. This allowed us to dynamically generate an interactive UI without parsing errors.
  2. Practice Mode: We implemented a custom spaced-repetition-style queue system on the frontend. Cards the user struggles with are pushed to the back of the queue, ensuring they are reviewed again before the session ends.
  3. Interactive AI Tutor: We used the Gemini Chat API with customized systemInstructions to act as an encouraging university professor. Instead of just giving away answers, it uses the Socratic method to guide students to the correct conclusions.

🚧 Challenges we faced

  1. Structured AI Outputs: Initially, getting an LLM to reliably output flashcards that could be rendered by a React component was difficult. The AI would occasionally output markdown or broken JSON. We overcame this by implementing the Type.OBJECT and Type.ARRAY schemas within the responseSchema configuration, guaranteeing 100% reliable data structures.
  2. State Management in Practice Mode: Designing a seamless UX for the flashcard practice mode required complex React state management to handle card flipping, tracking progress, and dynamically reordering the queue based on user feedback without causing jarring UI re-renders.
  3. Handling Multimodal Input: Formatting base64 image data correctly alongside text prompts to satisfy the Gemini API requirements took some debugging, but we successfully built a pipeline that seamlessly handles text, images, or both simultaneously.

🧠 What we learned

Building Nexus EduAI in under 12 hours was an incredible learning experience.

  • Prompt Engineering & Schemas: We learned how to constrain and guide Large Language Models using strict JSON schemas, moving beyond simple chatbots to building deterministic, AI-driven software logic.

  • The Math of Memory: While building the Practice Mode, we researched the science of learning. We learned about the Ebbinghaus Forgetting Curve, which can be modeled using the LaTeX equation:

    $$ R = e^{-\frac{t}{S}} $$

    Where $R$ is memory retention, $S$ is the relative strength of memory, and $t$ is time. This mathematical concept inspired our decision to build a queue system that forces immediate recall of failed cards, laying the groundwork for a future, more complex spaced-repetition algorithm.

  • Hackathon Collaboration: We learned how to scope a project effectively, prioritize the most impactful features (like the flashcard generator), and work seamlessly as a team under a strict deadline.

🚀 What's next for Nexus EduAI?

We plan to integrate the Gemini Live API so students can literally talk to their AI tutor via voice and stream their laptop cameras while solving math problems on paper in real-time. We also want to implement a persistent database to track the $R = e^{-\frac{t}{S}}$ forgetting curve over weeks and months to optimize mid-term and final exam prep!

Built With

Share this project:

Updates