Automations and AI Expert

Automations & AI Engineer (Contract, Per-Project) — QuantumGrwth

Type: 1099 contractor, project-based
Rate: $275–$900 per workflow, scoped per project based on complexity
Start: Within 30 days

We're an AI automation agency serving home service businesses — landscaping, HVAC, plumbing, electrical.
Our clients don't care about "AI."

They care about missed calls, slow bids, no-show rates, and cash flow. We build the systems that fix those.

What you'll build

Custom workflow automations, with and without AI, for front-of-house and back-office operations:

Voice agents for inbound and outbound calls (booking, qualification, follow-up)
Appointment booking flows across text, webform, and voice
Speed-to-lead systems that beat the 5-minute window
Task management systems that automatically send contracts, book onboarding meetings on client calendar, and assign and track project management tasks/activities between clients and our agency.
Financial reporting pipelines and recommendation engines
Bid and estimator agents, lead qualification, upsell, and cold outreach flows

Every build ships to a real business. There is no "demo mode."

Non-negotiable engineering standards

--> Every workflow you ship must be built with four things in mind, and you must be able to show us they existed in your past work:

Reliability. Retry logic, idempotency, dead-letter handling, graceful degradation when an LLM or API is down.
Sustainability. Documented, maintainable, not a 47-node spaghetti graph only you can read.
Security. Secrets management, PII handling, API scoping, auth done right.
Cost efficiency. Model routing, caching, batching, right-sized DBs, no $2K/month OpenAI bills for a $500 workflow.

You must also show real evidence of:

Evaluation frameworks for AI outputs (not vibes-based testing)
Governance and access setup (who can touch what, audit trails, versioning)
Deliberate database selection (why Postgres here, why a vector store there, why not Airtable)
Loop engineering (bounded, cost-controlled, exit conditions)
Error handling that a non-engineer client can survive

If your portfolio is "I made a cool automation," you are not the fit. If your portfolio is "here's how I d ---------- between two architectures and the tradeoff I picked," keep reading.

Required stack — shipped production work required in ALL of these
n8n (self-hosted or cloud, production workloads)
Claude Code / Anthropic API
OpenRouter or equivalent multi-LLM routing
Retell, Vapi, or comparable voice agent platform
Twilio or comparable messaging infrastructure
Supabase, Postgres, and vector DBs

"I did a tutorial" is not shipped production work. We will ask what broke and how you fixed it.

Minimum Requirements:
2+ years building production automations and AI workflows for real businesses
Public portfolio, GitHub, or client-facing case studies with proof of work (redact what you need to)
Written or diagrammed evidence of your architectural decisions in at least one prior build
-----------Please do not apply if you have ----------------------------
Only used ---------- or Zapier - Those are fine tools; they are not our stack.
No public portfolio, GitHub, or documented builds we can review.

Email with the subject line:

Automation Engineer Application — [Your Name] — Ready to Build

The first line of your email must answer this question: What is the single most expensive mistake you've made in a production automation, and what did you change in your process because of it?

Then include:

2–3 links to production builds you shipped (with your role clearly defined)
One redacted architecture diagram, Loom, or write-up showing how you made a reliability, cost, or security tradeoff
Your typical turnaround for a mid-complexity workflow (e.g., a voice agent that books appointments into a CRM and hands off to human on edge cases)
The stack items above where you have shipped production work, with one sentence each on what you built

Applications missing the subject line or the first-line answer will not be reviewed.

Thank you.