AITA - Sổ tay của Cô 📚

An AI math tutor assistant for Vietnamese teachers — speak or type a problem, get a clear step-by-step solution to share with your class.

Built for grades 6–9 math teachers. Voice-first, large text, high contrast.

Available on both desktop and mobile.

What it does

A teacher says or types a hard math problem. The app sends it to Gemini and returns a full Vietnamese explanation, step by step, ready to teach.

Two ways to ask:

• Voice — tap the green button, speak the problem out loud
• Text — type it in the box and hit Send

Tech stack

Part	Tech
Frontend	Angular 21, Tailwind CSS, Angular Material, KaTeX
Backend	NestJS 11, Prisma 5, Google Generative AI SDK
Auth	Google OAuth2 → JWT
Database	PostgreSQL 16

Run it locally

What you need

• Node.js 20+
• Docker (for Postgres)
• A Google Gemini API key
• A Google OAuth2 client — create one here
  • Add redirect URI: http://localhost:3000/auth/google/callback
  • Add JavaScript origin: http://localhost:4200

1. Start the database

docker run -d --name ai-ta-db \
  -e POSTGRES_USER=postgres \
  -e POSTGRES_PASSWORD=postgres \
  -e POSTGRES_DB=ai_ta \
  -p 5432:5432 \
  -v ai-ta-pgdata:/var/lib/postgresql/data \
  postgres:16

Manage the database

Start the database

docker start ai-ta-db

Stop the database

docker stop ai-ta-db

2. Set up the backend

cd service/ai-ta-api
npm install

Create service/ai-ta-api/.env:

GOOGLE_CLIENT_ID=your-client-id.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=your-client-secret
GOOGLE_CALLBACK_URL=http://localhost:3000/auth/google/callback

FRONTEND_URL=http://localhost:4200

JWT_SECRET=replace-with-any-long-random-string

API_KEY=your-gemini-api-key
DEFAULT_MODEL=gemini-2.0-flash

DATABASE_URL=postgresql://postgres:postgres@localhost:5432/ai_ta?schema=public

Apply the database schema:

npx prisma migrate deploy
npx prisma generate

Start the server:

npm run start:dev

Check it's running:

curl http://localhost:3000/health

Response:

{ 
  "status": "ok"
}

3. Set up the frontend

cd app/ai-ta-app
npm install
npm run start

4. Open the app

Go to http://localhost:4200 and sign in with Google.

Deploy

Service	Where
Backend	Render — render.yaml blueprint included
Frontend	Vercel — app/ai-ta-app/vercel.json included

Render: set API_KEY, GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET, FRONTEND_URL as environment secrets in the dashboard.

Vercel: set root directory to app/ai-ta-app, add API_URL env var pointing to your Render backend URL.

After deploying, update your Google OAuth client's redirect URI and JavaScript origins to match the live URLs.

Project layout

ai-ta/
├── app/ai-ta-app/        Angular frontend
│   ├── src/app/tutor/     main chat screen
│   ├── src/app/auth/      Google login + JWT handling
│   └── vercel.json        Vercel build config
├── service/ai-ta-api/    NestJS backend
│   ├── src/auth/          OAuth2 + JWT strategies
│   ├── src/gemini/        Gemini chat endpoint
│   ├── src/sessions/      chat history CRUD
│   ├── src/health/        health check endpoint
│   └── prisma/            schema + migrations
└── render.yaml           Render Blueprint (backend + db)