Case Study

Engineering Hub

Engineering Hub turns scattered course materials into an adaptive study system for engineering students. A PDF-to-questions pipeline powers 426 courses and ~142,000 practice questions, a Bayesian knowledge-tracing engine schedules what to review next, and the same account works on Android and the web, all served from the Cloudflare edge.

Built for BuildVerse 2026 · The John Amhanesi Foundation

Visit SiteGoogle PlayBack to Projects

Key Highlights

Downloads
300+ on Google Play
Platforms
Android app + Web app
Course Content
426 courses · 9 departments
Practice Content
~142,000 generated questions
Learning Engine
Bayesian Knowledge Tracing + spaced repetition
AI Integration
Google Gemini (quiz gen, marking, assistant)
Infrastructure
Cloudflare Workers + D1 + R2
Core Value
Adaptive learning that decides what you review next

Product Screenshots

Core Features

AI Study Companion

  • - Generates MCQs, theory questions, and fill-in-the-gap practice from course materials
  • - Marks theory answers with strengths, gaps, and practical improvement suggestions
  • - Supports image-based review for handwritten solutions
  • - Handles LaTeX-heavy math and engineering content
  • - Course assistant for quick explanations while studying

CourseGen Content Pipeline

  • - Turns raw textbook PDFs into structured courses: OCR, semantic embeddings (BGE-M3), ChromaDB retrieval, then Gemini-generated outlines and questions
  • - Produced 426 courses and ~142,000 questions across 9 engineering departments
  • - Resumable, cost-tracked, and key-balanced for high-volume generation

Adaptive Learning Engine

  • - Bayesian Knowledge Tracing estimates mastery probability for every topic
  • - Spaced repetition schedules each question and flashcard for the right review day
  • - The same engine runs in Dart and TypeScript, so phone and web stay in sync

Flashcards + Cross-Device Sync

  • - Flip-card flashcards with explanations
  • - Quiz attempts sync across devices, so a session started on the phone continues on the web
  • - Profiles persist to the cloud

Course & Content Management

  • - Courses grouped by department and academic level
  • - Local caching for faster repeat access and offline use
  • - In-app PDF viewing tuned for textbooks and lecture notes
  • - Course editor with a 10-course limit and review mode

Edge-Powered Document Delivery

  • - Cloudflare R2 storage with range requests for large PDFs
  • - Worker proxy and long-lived caching for faster document loads
  • - Browser-safe document access with controlled CORS

Progress + Study Habits

  • - Departmental timetables and study reminders
  • - Streaks and lightweight rewards for consistency
  • - Progress analytics that make improvement visible

Architecture & Stack

Frontend

  • - Flutter app (Clean Architecture, BLoC) for Android
  • - React 19 + Vite + TypeScript web app
  • - One adaptive engine implemented in Dart and TypeScript

API + Data

  • - Cloudflare Workers edge API
  • - D1 (SQLite) with 8 tables, 426 courses, ~142K questions
  • - Firebase Auth (email + Google)
  • - R2 for course documents and uploads

AI + Content

  • - Google Gemini for quiz generation, marking, and answers
  • - BGE-M3 embeddings + ChromaDB for RAG retrieval
  • - CourseGen Python pipeline (multi-engine OCR, resumable)

Want something like this built for your users?

If you’re building an AI-enabled product or a high-performance app, I can help with architecture, implementation, and shipping.

Contact MeBack Home