AI Observability

Full-stack AI observability app with:

• FastAPI backend (app/)
• React + Vite dashboard (dashboard/)
• SQLite by default (ai_logs.db)

Full Explaination of the Project

• Article 1
• Article 2

Features

• Register and login with role-based users (admin, employee)
• Send prompts to an LLM
• Track request cost and latency
• View request logs
  • admin: sees all logs
  • employee: sees only their own logs

Project Structure

app/
 ├── api/
 │    ├── admin_routes.py
 │    ├── auth_routes.py
 │    ├── employee_routes.py
 │    ├── __init__.py
 ├── core/
 │    ├── dependencies.py
 │    ├── security.py
 │    ├── __init__.py
 ├── models/
 │    ├── llm_log.py
 │    ├── user.py
 │    ├── __init__.py
 ├── services/
 │    ├── auth_service.py
 │    ├── llm_service.py
 │    ├── logging_service.py
 │    ├── __init__.py
 ├── config.py
 ├── database.py
 ├── main.py
 ├── schemas.py

Prerequisites

• Python 3.11+
• Node.js 18+ (or latest LTS)
• npm
• OpenAI API key

1) Backend Setup

From project root:

python -m venv .venv
.venv\Scripts\Activate.ps1
pip install --upgrade pip
pip install -r requirements.txt
pip install "python-jose[cryptography]" "passlib[bcrypt]"

Create/update .env in project root:

DATABASE_URL=sqlite:///./ai_logs.db
OPENAI_API_KEY=your_openai_api_key_here
SECRET_KEY=your_secret_key_here

Run backend:

uvicorn app.main:app --reload --host 127.0.0.1 --port 8000

Backend URL: http://127.0.0.1:8000

2) Frontend Setup

In a new terminal:

cd dashboard
npm install
npm run dev

Dashboard URL (usually): http://127.0.0.1:5173

3) First-Time Usage

1. Open the dashboard.
2. Register a user using API (Swagger/Postman/curl), then login in UI.
3. Submit prompts from the dashboard.

Open Swagger docs:

• http://127.0.0.1:8000/docs

Useful endpoints:

• POST /auth/register
• POST /auth/login
• POST /employee/generate (Bearer token required)
• GET /admin/logs (Bearer token required)

Example register payload:

{
  "name": "Admin User",
  "email": "admin@example.com",
  "password": "admin123",
  "role": "admin"
}

Employee payload is the same with "role": "employee".

Run with Docker (backend only)

From project root:

docker build -t ai-observability .
docker run --rm -p 8000:8000 --env-file .env ai-observability

Troubleshooting

• 403 Not authorized on generate/logs:
  • Login again to refresh token.
  • Ensure role is admin or employee.
• 401 Invalid token:
  • Token missing/expired, login again.
• OpenAI errors:
  • Check OPENAI_API_KEY in .env.
• Missing module errors (jose, passlib):
  • Install extra packages:
    • pip install "python-jose[cryptography]" "passlib[bcrypt]"

Notes

• CORS is currently open (allow_origins=["*"]) for local development.
• Default DB is SQLite file ai_logs.db.