Voice-Agent + Messaging Engineer — (Telnyx or similar)

The work Short, focused sprint: stand up, deploy, and hand off a service that reaches users over voice, WhatsApp, and Telegram. We already have a running Node scaffold (with the core state machine and a mock mode), the data contracts, and a telephony sandbox. We need a senior engineer to take it from "runs in mock on a laptop" to "deployed in a sandbox and working across all three channels," make the key stack/deployment calls, and onboard our in-house engineer to take over the full build. This is a setup + knowledge-transfer sprint, not a feature build — our engineer builds the features after you've laid the rails. What you'll do (the mechanics) -Recommend the stack for all three channels: voice layer (Telnyx vs Vapi / LiveKit / Retell / Twilio), WhatsApp (BSP / Cloud API), and Telegram (Bot API) — with rationale and trade-offs. -Configure the telephony sandbox: number, Call Control application, webhook configuration, credentials/secrets. -Deploy the scaffold to a sandbox host with a stable HTTPS webhook URL, beyond mock — so a real call connects and webhooks are received and routed. -Validate a real voice call on test numbers: an outbound call and/or an event-triggered two-way bridge with answering-machine detection and a cascade to a backup contact (never landing on voicemail). -Stand up the WhatsApp BSP sandbox and a Telegram bot: prove send + receive on each, and the quick-reply → action mapping pattern. -Set up the foundation: repo structure, a basic CI/deploy pipeline, secrets pattern, logging hooks, a minimal durable store, and a runbook — covering all three channels. -Onboard our in-house engineer: pairing/walkthrough sessions so he can run, deploy, and extend all three channels independently. Already provided (so you can move fast) A runnable Node scaffold (state machine + mock mode) · the data contracts · a telephony sandbox + test numbers · a sandbox hosting target · our in-house engineer available for pairing and to accept the work. Must-have - Shipped and deployed production voice agents / programmable telephony (Telnyx, Twilio, Vapi, LiveKit, Retell, or similar): call control, bridging, AMD, media streaming, TTS/STT. - WhatsApp Business (BSP / Cloud API) and Telegram Bot API integration. - Strong on deployment / DevOps for a Node service — HTTPS webhook endpoints, secrets, environments, CI/deploy. - Experience handing off / mentoring — you've set something up and made an in-house engineer productive on it. Clear communicator; tidy, documented, handoff-quality work. Nice-to-have Real-time / low-latency voice tuning · multi-language voice (incl. Spanish) · fleet / IoT or safety-critical comms. Tech We use Telnyx and are open to alternatives — tell us what you'd use and why. WhatsApp via a BSP / Cloud API; Telegram via the Bot API. Scaffold is Node/TypeScript. Engagement Fixed-price, ~1–2 weeks to set up and hand off all three channels (voice, WhatsApp, Telegram). Full scope, tasks, acceptance, and the deadline are in a Statement of Work we'll share with shortlisted candidates (under NDA). You work against sandbox/stubbed endpoints; our in-house engineer wires the internal services and signs off. To apply, please briefly answer with a loom - Walk me through how you'd deploy a Node telephony service with a stable webhook endpoint and validate a live call in a sandbox — host, webhook URL, secrets — and whether you'd use Telnyx or an alternative. - An event triggers a call and the primary contact doesn't answer within ~18 seconds. How do you cascade to a backup and ensure you never connect to voicemail? - WhatsApp + Telegram: which BSP/approach would you use for each, and one gotcha you've hit? - Tell me about a time you set something up and handed it to an in-house engineer — how did you make sure they could run with it?