LiveKit Voice AI Setup

I need LiveKit running on an AWS domestic (India) server so I can power a Voice AI experience. The core use-case is “LiveKit Voice AI”: users will place and receive AI-driven voice calls inside our app, in real time, with low latency inside India.

Key feature requirements
• AI voice calling enabled through LiveKit
• Automated call handling so calls can be routed or answered without a human agent
• Speech-to-text transcription in real time, ready for downstream processing
• Call analytics captured and stored for later insight

Scope of work
– Provision and secure an AWS account or VPC inside an India region (Mumbai preferred)
– Deploy and configure LiveKit server components, making sure TURN / STUN services work reliably for the local network conditions
– Integrate the Voice AI layer so calls can be automatically answered, transcribed, and analyzed
– Expose clean APIs or WebSocket endpoints that my application can call to start, monitor, and terminate AI voice sessions
– Provide basic monitoring and logging so we can trace call quality and server health
-- Integrating SIP trunking of an Indian number to Livekit server

Acceptance criteria
1. A test call placed through the API is automatically picked up by the AI and transcribed with <1 s delay.
2. Call metadata (duration, participant IDs, sentiment or other analytics) is written to a database or S3 bucket.
3. All traffic stays within AWS India regions for compliance.
4. Complete deployment scripts (Terraform, CloudFormation, or Docker Compose) are handed over so I can recreate the stack.

Please tell me which LiveKit version you would deploy, the AWS services you expect to use, and an estimated timeline for getting a first working demo online.