Welcome to our project for HackAI 2026, Locked In!

This platform redefines the university study experience by leveraging Large Language Models to automatically organize a student's entire semester and gamifying the learning process through a live, competitive matchmaking system.

The Vision

University students waste countless hours navigating half a dozen disorderly PDF syllabi and class outlines to figure out what they need to study and when.

Our platform solves this by:

1. Instantly parsing every syllabus a student uploads, as well as directly pulling syllabi from Nebula to parse.
2. Generating a master chronological timeline of deadlines.
3. Automatically querying Nebula APIs to assign difficulty ratings based on historical grade distributions.
4. Using Gemini 2.5 Pro to create a granular, day-by-day study plan.

And the best part? Students can queue up in a live, Ranked Matchmaking system against other students taking the exact same class, competing in AI-generated quizzes over the exact material they are currently learning in lecture that week.

Architecture

This is a monolithic repository containing two main segments:

/api (Backend)

A high-performance Python FastAPI application connected to MongoDB, hosted on MongoDB Atlas.

• Authenticates users securely via Clerk JWTs.
• Features an abstracted GeminiClient utilizing both gemini-2.5-flash (for fast data extraction and quiz generation) and gemini-2.5-pro (for deep study-plan reasoning).
• Houses the entire HTTP-polling Multiplayer Matchmaking engine, complete with live status tracking, cross-referenced syllabus topic generation, and a localized per-class Elo system.
• See /api/README.md for local setup instructions!

/webapp/clerk-react (Frontend)

A modern, responsive Single Page Application built with React, Vite, and Tailwind CSS.

• Leverages Clerk for rapid, secure user authentication.
• Parses PDF and DOCX files entirely client-side (using pdf.js and mammoth) before sending pure text to the backend, drastically reducing server costs.
• Dynamically renders chronological timelines, daily study plans, and live global/local Elo leaderboards using React Router.
• Features an isolated Matchmaking Arena (/#match) where users can queue, view live poll updates, take their AI-generated quizzes, and watch their rank go up in real time!

Getting Started

To run the full stack locally, you will need two terminal windows:

Terminal 1 (Backend API)

cd api
source venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload

Terminal 2 (Frontend React App)

cd webapp/clerk-react
npm install
npm run dev

(Ensure you have populated the .env files in both directories according to their respective requirements!)