Deploy a finished Node.js AI-voice + CRM app to production (code complete — deploy, wire APIs, test)

We run a non-invasive body-sculpting & skin clinic in Perth, Western Australia. We have a
fully written, code-complete AI phone-receptionist + CRM (Node.js). This is not a
build-from-scratch job — the software exists, is documented, and passes validation. We need an
experienced developer to deploy it, connect our API accounts, wire the webhooks, and prove it
works end-to-end, then hand it back for us to run.

You will not be writing the product or the ad campaigns. You're the person who takes a
finished codebase and stands it up correctly and safely in production.

WHAT YOU'LL DO


Deploy the provided app to an always-on host (Render / Fly / Railway / a VPS — your
recommendation). It's a persistent Node service with a WebSocket and a small native SQLite
dependency; it ships with a Dockerfile and a Render blueprint.
Configure environment variables from the keys we provide, and generate the three app
secrets (npm run secrets). Set PUBLIC_BASE_URL to the live HTTPS address (the voice call
builds its connection from this — it must be exact).
Wire the integrations:

Twilio ConversationRelay (voice) + SMS, using our number; confirm the ElevenLabs voice
is speaking via ConversationRelay (ttsProvider="ElevenLabs" + our voice ID).
Stripe — add the webhook endpoint (/webhook/stripe, checkout.session.completed) and
set the signing secret.
Anthropic key for the AI.



WordPress — install the provided mu-plugin "bridge" + the smart-landing plugin, and point
the lead webhook at the backend with the shared secret (forwarding name, mobile, and ad click
IDs). No theme/site building — just the plugin + hook.
One full end-to-end test pass: submit a test lead → AI calls → books against the built-in
scheduler → texts a Stripe test deposit → payment confirms the booking → reminders schedule.
Also verify the safety-screen path and the SMS STOP opt-out. Then switch Stripe to live.
Hand back delegated access + a short "what I did / how to restart it" note.


WHAT WE PROVIDE


The complete backend (hourglass-ai-backend), with README.md and an ordered SETUP.md
deploy checklist.
The WordPress smart-landing plugin + the lead-bridge plugin.
Our four API accounts' keys (Anthropic, Twilio, Stripe, ElevenLabs) and a keys sheet.
A plain-English overview of what the system does.


WHAT THIS ROLE IS NOT

No feature development (it's built), no ad campaigns, no landing-page design, no ongoing
full-service management. It's a deploy-configure-test-handover job, with optional light
maintenance if you want it.

REQUIRED SKILLS


Strong Node.js and deploying persistent services with WebSockets to production.
Comfortable with Docker, environment variables, and hosts like Render/Fly/Railway/VPS
(including a natively-compiled dependency — better-sqlite3 — which needs standard build tools
or a prebuilt binary).
Twilio (Programmable Voice/SMS; ConversationRelay a big plus), Stripe webhooks, and
basic WordPress plugin install.
Security-minded: secrets handling, keeping a database off the public web host, HTTPS.


NICE TO HAVE


Twilio ConversationRelay / voice-AI experience.
Awareness of handling sensitive/health data and call-recording compliance (AU).

TO APPLY — PLEASE ANSWER (screening)


Link two persistent Node apps you've deployed to production (ideally with a WebSocket or
telephony). Where were they hosted, and why there?
Our app has a small native dependency (better-sqlite3) and needs an always-on host
(no cold sleeps). How would you host it, and how do you handle the native build?
PUBLIC_BASE_URL must exactly match the live address or the voice call fails silently. How do
you verify the whole call path is working before we point real customers at it?
How do you keep a SQLite database of health information private and backed up, off the
public web host?
Pricing — split it two ways: (a) a one-off fee to deploy, wire, and test to a verified
working launch; and (b) an optional light monthly rate for monitoring/updates. Say what
each includes.