UniStudyHub

Overview

UniStudyHub is an advanced educational ecosystem built for university students. It provides a structured hierarchy containing rich notes, dynamic quizzes, and an integrated AI tutor. The platform supports heavy data ingestion through Python-based OCR for digitizing physical past questions and CSV bulk uploads for rapid curriculum onboarding.

The Problem

University study materials are often scattered, and STEM students specifically struggle with platforms that cannot render complex mathematical notation correctly. Furthermore, educators faced a massive bottleneck trying to digitize decades of physical past exam papers and manually input hundreds of student records.

Process

I built the core infrastructure with Next.js and Supabase. To solve the data-entry bottleneck, I integrated Python-based OCR workflows to automatically extract and structure text from scanned past questions, alongside a robust CSV upload system for bulk-importing quizzes and users. For STEM subjects, I integrated MathVisual to perfectly render complex mathematical constraints directly in the browser.

Technical Deep-Dive

Building the multi-modal data ingestion pipeline was a massive challenge—specifically ensuring the OCR extraction accurately preserved mathematical formulas before being passed to MathVisual for frontend rendering, and validating large CSV payloads gracefully. On the gamification side, I designed a normalized ranking engine for the Global Leaderboard that calculates percentage accuracy and speed-per-question to fairly rank quizzes of wildly varying lengths.

Code Sample

Normalized ranking algorithm — fairly ranks quizzes of different lengths by accuracy percentage and speed per question.

// lib/leaderboard.ts
function rankEntries(raw: LeaderboardEntry[]) {
  // Sort by Accuracy (Percentage) first, then Speed (Time per question)
  const sorted = [...raw].sort((a, b) => {
    const pctA = a.max_score > 0 ? a.score / a.max_score : 0;
    const pctB = b.max_score > 0 ? b.score / b.max_score : 0;
    if (pctA !== pctB) return pctB - pctA;

    const timePerQuestionA = a.max_score > 0 ? a.time_taken_seconds / a.max_score : Infinity;
    const timePerQuestionB = b.max_score > 0 ? b.time_taken_seconds / b.max_score : Infinity;
    return timePerQuestionA - timePerQuestionB;
  });

  let pos = 1;
  return sorted.map((entry, i) => {
    if (i > 0) {
      const prev = sorted[i - 1];
      const prevPct = prev.max_score > 0 ? prev.score / prev.max_score : 0;
      const currPct = entry.max_score > 0 ? entry.score / entry.max_score : 0;
      const prevTPQ = prev.max_score > 0 ? prev.time_taken_seconds / prev.max_score : Infinity;
      const currTPQ = entry.max_score > 0 ? entry.time_taken_seconds / entry.max_score : Infinity;

      // Only increment position if the accuracy or speed actually differs
      if (currPct !== prevPct || currTPQ !== prevTPQ) pos = i + 1;
    }
    return { ...entry, rank: pos };
  });
}

Results