Production Voice AI: SIP Telephony LiveKit AWS Docker Python

Production-Ready Voice AI Agents with SIP Telephony (+91), LiveKit, AWS EC2, and Docker for Inbound & Outbound Calls

What you’ll learn
✓ Design a production-grade Voice AI architecture integrating PSTN, SIP, LiveKit, and AWS.
✓ Configure inbound and outbound SIP trunks with a real telecom provider.
✓ Configure a real Indian DID number using VoBiz AI (Indian telecom provider).
✓ Deploy a Voice AI agent on AWS EC2 (Ubuntu) using Docker containers.
✓ Implement a complete STT → LLM → TTS pipeline for real-time voice interactions.
✓ Build and test an inbound AI Complaint Board system connected to a real DID number.
✓ Develop and trigger an outbound AI calling system for automated call campaigns.
✓ Integrate LiveKit for real-time media streaming and SIP call routing.
✓ Configure API integrations including Deepgram, OpenAI, Cartesia, and Sarvam TTS.
✓ Optimize voice responses by modifying TTS pace, pitch, and performance settings.
✓ Analyze scalability, concurrency limits, and production performance of Voice AI systems.

Requirements
● Basic understanding of Python programming
● Familiarity with Linux/Ubuntu command line
● Basic knowledge of cloud computing (AWS fundamentals)
● Understanding of APIs and environment variables
● A computer capable of running Docker and SSH tools (like MobaXterm or Terminal)

Description
In This Course, You Will Build

In this course, you will design and deploy a complete production-grade Voice AI system, including

• A live Inbound AI Complaint Board that answers real phone calls (+91, India Number)

• A real Outbound AI Calling System for an Eye Hospital

• A centralized Voice AI Dashboard to monitor and manage call activity

• A fully deployed Voice AI agent on AWS EC2 (Ubuntu)

• A containerized architecture using Docker

• End-to-end SIP trunk integration with a real telecom provider

• Live media routing using LiveKit

• A complete STT → LLM → TTS pipeline for real-time voice intelligence

• Scalable infrastructure capable of handling concurrent calls

By the end of this course, you will have a fully functional, telecom-integrated AI voice system that can answer and make real inbound and outbound phone calls in a production environment.

Course Curriculum

Course Introduction: Live Voice AI System

• Live Demo – Complaint Board Inbound

• Live Demo – Eye Hospital Outbound

• Live Demo – Voice AI Dashboard

• What You Will Build in This Course

• What Makes This Voice AI Course Different

• End-to-End Call Flow

Voice AI and SIP Telephony Fundamentals

• What You Will Learn in This Module

• Understanding PSTN (Public Switched Telephone Network)

• What Is SIP (Session Initiation Protocol)?

• PSTN vs SIP

• Inbound Call Flow

• Outbound Call Flow

• SIP Trunk in Inbound Call Flow

• SIP Trunk in Outbound Call Flow

• Understanding SIP Credentials

• How Production AI Voice Systems Work

AWS Infrastructure and Server Setup for a Voice AI Agent

• What You Will Learn in This Module

• Ubuntu EC2 Server (Voice AI Agent Server)

• Creating an Ubuntu EC2 Instance

• Accessing the Server via MobaXterm

• Ubuntu AI Server (STT => LLM => TTS)

• Docker Setup for Voice AI Agent

Developing the Voice AI Agent Code

• Overview of Voice AI Agent Code

• Deep Dive into the Voice AI Agent Script – Part 1

• Deep Dive into the Voice AI Agent Script – Part 2

• Deep Dive into the Voice AI Agent Script – Part 3

• Deep Dive into the Voice AI Agent Script – Part 4

Setting Up API Keys (STT → LLM → TTS)

• Deepgram, OpenAI, and Cartesia API Keys

Why Do We Need Livekit

• What is Livekit

• Why We Use Livekit for Our AI Project

• LiveKit Documentation Overview

• What Problem Does Livekit Solve?

• Setting Up LiveKit Account and APIs

Configuring Prompts, Dependencies, and Dockerfile

• Developing prompts for the Voice AI Agent

• Required Python Dependencies

• Building the Dockerfile for the Voice AI Agent

Run the Voice AI Agent in a Docker Container

• Build the Docker Image Using a Dockerfile

• Launch the Voice AI Agent in a Docker Container

• Test the Voice AI Agent Using LiveKit Playground

Sarvam AI TTS (Text-to-Speech) for the Voice AI Agent

• Sarvam AI Bulbul V3 TTS: Overview

• Integrate Sarvam TTS into Our Voice Agent

• Add the Sarvam ai Plugin for LiveKit Agents

• Recreate the Docker Image of the Voice Agent

• Relaunch the Voice AI Agent in a Docker Container

• Add the Sarvam API and Test the New Voice

Connecting AI Voice Agents Using SIP

• SIP-Based Integration of AI Voice Agents

• SIP Provider Selection

VoBiz AI Account Setup & DID Provisioning (Telecom Provider)

• Create an Account in VoBiz AI (Telecom Provider)

• Purchase a DID Number

• VoBiz AI DID supports 10 concurrent calls

Inbound SIP Trunk at the Telecom Layer

• Create a Inbound Trunk

• Create an Origination URI

• Link the DID Number to the Inbound Trunk

• Add Agent Name in EntryPoint (agent,py)

• Rebuild and Relaunch the Container

Inbound SIP Trunk Configuration in LiveKit

• LiveKit: Inbound SIP Trunk Setup

• Create a Inbound Trunk

• Dial the DID Number and Test the Voice AI Agent

Building an Outbound Voice AI System for an Eye Hospital

• Outbound Call Flow – Complete Overview

• Create Outbound SIP Trunk

• Create an Outbound Trunk in LiveKit

• Project Overview – Outbound AI Calling System

• Agent Creation – agent,py

• Project File Setup – Prompts & Dependencies

• Dockerfile Setup

• Build Voice AI Agent Docker Image

• Run Voice AI Agent Container

• Create Python Environment

• Install Python Dependencies

• Trigger Outbound Call

Voice Optimization and Final Call Testing

• Modifying TTS Pace and Pitch

• Re-Trigger the Outbound Call

Call Handling Capacity and Scalability

• What You Will Learn in This Module

• System Architecture Overview

• Vobiz AI Channel Allocation (Telecom Provider)

• EC2 Instance Configuration – c6i.xlarge

• Real-Time Testing Results

• Theroretical vs Practical Limit

• When Performance May Degrade

• Last Lecture

Who this course is for
■ AI and Backend Developers who want to build real phone-based AI systems
■ DevOps Engineers deploying AI on AWS with Docker
■ Telecom / VoIP professionals exploring AI call automation
■ Startup founders building AI-powered calling solutions

https://anonymz.com/?https://www.udemy.com/course/livekit-voice-ai-agent