To deploy LiveKit agents on a FastAPI backend, you can follow these general steps: 1. Create a

Dockerfile

for your FastAPI application that ends with a command like

CMD ["python", "main.py", "start"]

. 2. Deploy the application to a platform that supports container orchestration, such as Kubernetes or Render.com. LiveKit agents are designed to work well in such environments. For example, in the avatar_agents setup, a FastAPI application serves as a dispatcher for avatar workers, with endpoints for launching workers in specific rooms. The dispatcher uses a

WorkerLauncher

class to manage worker processes. Additionally, LiveKit agents use WebSocket connections to register with the LiveKit server and accept incoming jobs. They do not require exposing inbound ports to the public internet, but you may optionally expose a private health check endpoint (default: http://0.0.0.0:8081/). If you need further assistance or have specific requirements, feel free to share more details! Sources: Deploying to production | LiveKit Docs