AI Chat Backend

Main Technologies Used

• Node.js (v22.x)
• TypeScript
• pnpm (package manager)
• PostgreSQL
• Ollama (LLM integration)
• Docker & Docker Compose
• Express.js
• Prisma ORM
• ESLint & Prettier (code quality) (Prettier)
• nginx - Used in Production Docker container
• Vitest

---

Getting Started

1. Clone the repository

git clone <repo-url>
cd ai-chat-backend

2. Install dependencies

Make sure you have pnpm installed:

pnpm install

3. Set up environment variables

Create a .env file in the root directory with the following variables:

APP_PORT=<port that will used to start the server>
FRONTEND_ALLOWED_URL=<frontend app allowed url>

OLLAMA_HOST=<host where ollama server will run>
OLLAMA_PORT=<port where ollama server will run>
MODEL_ID=<ollama llm model id>

DB_NAME=<database name>
DB_USER=<database user name>
DB_PASSWORD=<database user password>
DB_PORT=<database port>
DATABASE_URL=<database connection string for ORM>

Adjust values as needed for your setup.

By default the values are:

APP_PORT=3000
FRONTEND_ALLOWED_URL=http://localhost:5173

OLLAMA_HOST=ollama
OLLAMA_PORT=11434
MODEL_ID=llama3.2:latest

DB_NAME=ai_chat
DB_USER=postgres
DB_PASSWORD=12345
DB_PORT=5432
DATABASE_URL="postgresql://postgres:12345@postgres-db:5432/ai_chat?schema=public"

4. Start the Ollama server

Make sure you have Ollama installed locally. In a separate terminal, run:

ollama serve

This will start the Ollama LLM server required by the backend.

5. Set up the database

Make sure you have PostgreSQL running and the credentials match your .env file. Run migrations and generate Prisma client:

pnpm db:migrate:dev

6. Seed the database

pnpm db:seed

7. Start the application

pnpm dev

The server will start on http://localhost:3000 by default.

7. Run Prisma Studio Web UI

pnpm prisma-studio

8. Run tests

pnpm test

or

pnpm test:watch

---

Running with Docker in Development mode

1. Build and start all services

pnpm docker:dev

This will start:

• The backend app
• PostgreSQL database
• Ollama LLM service
• Prisma Studio Web UI

Ports will be gotten from .env file

2. Environment variables

Docker Compose will use the .env file in the root directory. Make sure it is set up as described above.

3. Access

• API: http://localhost:3000/api
• Health check: http://localhost:3000/api/health-check

---

Running with Docker in Production mode

1. Build and start all services

pnpm docker:prod

This will start:

• nginx server gateway
• The backend app
• PostgreSQL database
• Ollama LLM service

Ports will be gotten from .env file

2. Environment variables

Docker Compose will use the .env file in the root directory. Make sure it is set up as described above.

3. Access

• API: http://localhost:80/api
• Health check: http://localhost:80/api/health-check

---

Useful Scripts

• pnpm start - Start the backend in production mode
• pnpm dev — Start the backend in development mode
• pnpm db:migrate:dev — Run database migrations and generate Prisma client in Development environment
• pnpm db:migrate:prod - Run database migrations and generate Prisma client in Production environment
• pnpm db:seed - Run seed script to populate the database
• pnpm lint — Lint the codebase
• pnpm format — Format the codebase
• pnpm prisma-studio - Run Prisma Studio Web UI to inspect models in the database
• pnpm docker:dev - Run development Docker container
• pnpm docker:prod - Run production Docker container
• pnpm test - Run tests once
• pnpm test:watch - Run tests in the watch mode
• pnpm test:coverage - Display test coverage
• pnpm test:ui - Run vitest ui

---