Build a Voice Platform Core

Build a Voice Platform Core — Asterisk + FreeSWITCH with a Clean Call-Control API This project is to build a telephony core end to end: stand up the voice engines, then build the API that applications connect through. 1) The voice core — deploy & configure the engines - Deploy, configure and harden Asterisk (production) and FreeSWITCH (second engine) - SBC (Kamailio) — SIP/TLS, SIP-trunk/carrier connectivity, routing, fraud control - Media (rtpengine) — RTP/SRTP, transcoding, recording - Multi-tenant and highly available 2) The engine-control API — the core primitives only A clean REST + real-time webhook/WebSocket API exposing the irreducible call-control primitives: Originate · Answer · Hangup · Bridge · Transfer (blind/warm) · Hold/Resume · Mute · Conference · Play audio · Collect & Send DTMF · Record (start/stop/pause) · Monitor/Whisper/Barge · Set caller-ID · Answering-machine detection · Call-state · Call event stream · CDR access Build only the primitives. Anything that is merely a combination of them — IVR, callback, number masking, dialer/campaign, voicemail — is out of scope and built separately on top. Do not build those. Engine independence is the key requirement. The API must not leak engine internals. It must be proven by exposing the same API over both Asterisk and FreeSWITCH with identical behaviour, validated in shadow mode — so the engine can be swapped without anything that calls the API noticing. Also required: a durable event pipeline (Kafka) with retry/replay so no call event is lost, Kubernetes + IaC delivery, HA failover, SRTP/mTLS, full observability. Verified by an automated test suite passing on both engines, covering media-path and timing — not just control-plane. Must have: deep Asterisk (ARI/AMI) AND FreeSWITCH (ESL), Kamailio/SBC + SIP trunking, rtpengine, REST/WebSocket API design, Kafka/event pipelines, Kubernetes.