MultiAgent Backend is a FastAPI‑based service that exposes two AI agents—Support Agent and Dashboard Agent—through a versioned REST API.
It integrates CrewAI to coordinate multi-agent workflows, uses MongoDB Atlas for structured data persistence, and leverages the Gemini API for LLM-powered prompt processing.
This backend is production‑ready, configured for Gunicorn + Uvicorn behind NGINX, and secured with HTTPS via Certbot.
-
Deployed on AWS EC2 with nginx
-
API versioning under
/api/v1 -
Two independent AI agents using CREW AI with Gemini API
• Support Agent for external client queries
• Dashboard Agent for internal analytics -
Health‑check endpoint at
/health -
CORS enabled
-
MongoDB Atlas integration
-
Environment‑variable support via
dotenv -
Deployment‑ready with Gunicorn workers and NGINX reverse proxy
Access the full API documentation and sample prompt usage here:
🔗 https://documenter.getpostman.com/view/28016254/2sB34co2ZD
multiagent-backend/
│
├─ src/
│ ├─ config/ # Agent and DB configuration
│ ├─ routes/ # FastAPI routers
│ ├─ tools/ # Agent tools
│ └─ main.py # FastAPI app entry point
│
├─ .env # Environment variables (git‑ignored)
├─ requirements.txt # Python dependencies
└─ README.mdx # Documentation
git clone https://github.com/your-username/multiagent-backend.git
cd multiagent-backendpython3 -m venv venv
source venv/bin/activatepip install -r requirements.txttouch .env#Backend env
MONGO_CONNECTION_URL = your url
MONGO_DB_NAME= your db name
MODEL= your model
LITELLM_PROVIDER= your provider
GEMINI_API_KEY= your api key
#frontend env
NEXT_PUBLIC_API_URL= your api url (localhost one)
PYTHONPATH=src uvicorn main:app --host 0.0.0.0 --port 8000 --reloadThis project is licensed under the MIT License.
You are free to use, modify, and distribute this software with proper attribution.
For issues, bugs, or feature requests, please open an issue on the GitHub Issues page.
Aayush Tirmanwar
Computer Science Student | Full Stack Developer
Built with purpose, speed, and flexibility in mind.