Skip to content

Umer2000/EduAccessAI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 

Repository files navigation

🎓 EduAccess AI

Transforms classroom materials into dyslexia-friendly and screen-reader-ready content using Gemini + Opus.

🚀 Overview

EduAccess AI is an accessibility-focused platform that transforms traditional classroom materials into inclusive, student-ready learning experiences. Teachers upload any PDF or worksheet, and the system automatically:

Extracts and simplifies the content

Adapts it for dyslexic or blind learners

Generates reading levels

Passes the lesson through an Opus agentic review workflow

Updates lesson status and audit logs via webhook

Publishes approved lessons to a student dashboard

Built for the AI Genesis Hackathon, EduAccess AI delivers a seamless accessibility pipeline powered by Gemini 2.5 Flash and Opus.

✨ Features

👩‍🏫 Teacher Dashboard

Upload PDFs / images

Run AI-powered content simplification

Trigger Opus workflows for automated review

View lesson audit reports and approval status

Manage lessons across profiles and languages

🧑‍🎓 Student Dashboard

Browse teacher-approved lessons

Filter by profile & language

Read simplified accessible text

Listen using built-in text-to-speech

View bullet-point summaries & reading level

⚙️ Backend (FastAPI)

PDF text extraction using pypdf

Gemini 2.5 Flash for simplification & summarization

AI-driven reading-level estimation

Lesson database stored in-memory (hackathon mode)

Accepts Opus webhook for final audit & approval

🤖 Opus Workflow (Fully Integrated)

The Opus pipeline includes:

1.API Trigger using inputs:

lesson_id, teacher_id, profile, language

2.Fetch Lesson node Fetches lesson content from FastAPI backend.

3.Agentic AI Review Gemini evaluates clarity, accessibility issues, structure, and reading level. Generates:

decision (ready_for_students or needs_changes)

issues list

overall comments

suggested reading level

4.Optional Human Reviewer Teacher or moderator can approve or reject.

5.Webhook Delivery Opus POSTs results to: /api/lessons/{lesson_id}/opus-webhook

FastAPI updates the lesson status, audit trail, and metadata.

6.Student Dashboard Updates Automatically Students instantly see approved content.

🧠 Architecture

Frontend (React + Vite + Vercel) ↓ FastAPI Backend (Render) ↓ Gemini 2.5 Flash (AI) ↓ Opus Workflow ↓ Webhook → FastAPI → Lesson Updated → Students See It

🛠️ Tech Stack

1.Frontend

React + Vite

TypeScript

Custom CSS

Deployed on Vercel

2.Backend

FastAPI + Uvicorn

Python

PDF extraction (pypdf)

Deployed on Render

3.AI

Gemini 2.5 Flash (simplification, summarization, language adaptation)

4.Automation

Opus Workflow Engine

API trigger

Fetch content

Agentic decisioning

Automated quality checks

Webhook callback

🔧 Setup Guide

1️⃣ Clone the repository

git clone https://github.com/Umer2000/EduAccessAI.git cd EduAccessAI

⚙️ Backend Setup (FastAPI)

.env example GEMINI_API_KEY=your_gemini_key GEMINI_MODEL_ID=gemini-2.5-flash OPUS_API_URL=https://api.opus.ai/workflows//start OPUS_API_KEY=your_opus_api_key

Install dependencies pip install -r requirements.txt

Run backend locally uvicorn main:app --reload

Backend runs at: http://127.0.0.1:8000

🖥️ Frontend Setup (Vite + React)

Go to frontend folder cd frontend Create .env VITE_API_BASE_URL=http://127.0.0.1:8000

Install and Run npm install npm run dev

☁️ Deployment

🟣 Backend → Render

Build: pip install -r requirements.txt Start: uvicorn main:app --host 0.0.0.0 --port $PORT

Add environment variables

🟢 Frontend → Vercel

Import GitHub repo

Add Vercel env var: VITE_API_BASE_URL=https://eduaccessai.onrender.com/ Build: npm run build Output directory: dist/

📌 Current Status

✔ Teacher dashboard

✔ Student dashboard

✔ Gemini 2.5 Flash integration

✔ Real-time PDF processing

✔ Language support: English + Spanish

✔ Live backend (Render)

✔ Live frontend (Vercel)

✔ Opus workflow fully integrated

✔ Webhook updating lessons correctly

✔ Full end-to-end accessibility pipeline

Roadmap

🔵 Add OCR for images-only PDFs

🔵 Add more languages

🔵 Add real database (Supabase / Firebase)

🔵 Add student analytics & reading progress

🔵Multi-teacher multi-classroom support

🔵 Improve audio using advanced TTS

👤 Author

Team: Genesis Hackers

Umer Anis- AI Developer

Shadab Ahmed-Accessibility Advocate

Built for the AI Genesis Hackathon(lablab.ai)

About

EduAccess AI transforms classroom materials into accessible, simplified versions for dyslexic and blind students

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors