Skip to content

MR-GREEN1337/ustadh

Repository files navigation

Here’s a refined version of your project README or planning document with the Loom iframe embedded properly and some formatting improvements for clarity and presentation:


🧑‍🏫 Ustadh — AI-powered LMS

Ustadh App Preview


🔁 Sprint Notes (11/05)

  • Implement course generator
  • Add preview & export for course sessions
  • Fix file uploads and finalize chat/session tools
  • Reduce complexity
  • Fix endpoint errors

🔄 Changelog

  • ✅ Switched from await fetch to useSWR
  • ✅ Added CAPTCHA to auth and all unauthenticated forms
  • ✅ Started building the blog

📹 Loom Preview

<iframe src="https://www.loom.com/embed/47ed91e3e5534c44b6263a993da1f4c3?sid=5edad967-877f-4bbf-b765-6ef32064dacf" frameborder="0" allowfullscreen style="width: 100%; max-width: 800px; height: 450px;"> </iframe>

🚧 Roadmap

✅ Core Features

  • CAPTCHA in auth
  • Amazigh language support (i18n)
  • Block student browsing during exams
  • Redirect logged-in users from landing → dashboard
  • Schedule management via image → JSON or manual input
  • CSRF protection
  • Store JWT in httpOnly cookie

🏫 School Onboarding

  • Role-based UI & navigation
  • Update user via AuthProvider
  • Analytics tables
  • Community + messaging services
  • Tutoring with Ibn Battuta branding
  • Admin & Professor dashboard features

✨ User Features

  • Real-time notes (WebSocket)
  • EditorJS for note-taking
  • Prompt templates by education level
  • Quizzes
  • “Report a Bug” sidebar link
  • Student performance analysis & suggestions

📅 Sprint (19/04 → 26/04)

  • Finalize community + tutoring
  • Admin/Professor UI enhancements
  • Implement AI RAG data layer
  • Upload curated Moroccan school data

🔄 Sync Feature

  • Synchronize avec l’école: Fetch updated schedules

💬 Messaging Logic

  • Students see only relevant teachers
  • Admins & Professors can message all users

🛠 Deployment Strategy

Stack Overview

Component Tech
Backend FastAPI (Cloud Run)
Clients React Native / Next.js (Vercel)
Database Neon (PostgreSQL)
Cache/Queue Upstash Redis
Worker Celery (Cloud Run Jobs)
Email SendGrid / Mailgun
Storage GCP Cloud Storage (optional)

Steps

  1. Backend: Dockerize + deploy to Cloud Run
  2. Frontend: Web (Vercel/Cloud Run), Mobile (HTTPS)
  3. Database: Neon w/ pooling + GCP Secrets
  4. Worker: Celery job via Cloud Run
  5. Email: Triggered via GCP Function or Pub/Sub
  6. Secrets: All stored in GCP Secret Manager

📊 Monitoring & Budget

  • Cloud Monitoring & Logging
  • Alerts: errors, cost, Redis usage
  • Free tiers for most services
  • Target: ~$20–40/month (up to 5,000 users)

🔮 Coming in May

  • Launch AgentX (AI-first features)
  • Complete UI overhaul
  • Refactor int IDs → UUIDs
  • Add modern floating header

Would you like this content exported to a Markdown file or used in a GitHub README?