AI Agent Engineer (Python) — Long-Term Role with Growing US Coaching Business

About Us

Growth Cleaning is a US-based coaching company helping professionals build remote cleaning businesses in the US and Canada. We coach 100+ students. We're scaling fast and need to automate everything with AI agents.

We already have a small remote team — including teammates in the Philippines — and we treat people well. This is a long-term role, not a project.

What You'll Do

You'll build and maintain Python AI agents that connect to real business tools (Close CRM, GoHighLevel, Stripe, QuickBooks, Slack, Supabase). These aren't chatbots — they're agents that pull data, make decisions using Claude and OpenAI APIs, and write results back to our systems on a schedule.

Examples of what we already have running and what you'll build more of:

Sales call grader that reads transcripts and scores closers
Bookkeeping agent that syncs Stripe and bank data to QuickBooks
Lead attribution agent that tracks where every booked call came from
Content pipeline agents (Instagram, YouTube)
Client dashboards (we sell AI agents to our coaching clients too)
Required Skills

3+ years Python (must be strong — we will test you)
Real experience using Claude API or OpenAI API in production code (tool use, structured outputs)
REST API integration (OAuth, webhooks, pagination)
SQL and Postgres / Supabase
Git and GitHub
Strong written English (you'll work with me directly on Slack)
Nice to Have

Next.js + Tailwind
Experience with Close CRM, GoHighLevel, QuickBooks, or Stripe APIs
MCP (Model Context Protocol) server experience
What We DON'T Want

"AI experts" who only know n8n / Make / Zapier — those are not coding
Anyone who has never written Python in production
Anyone who needs hand-holding on every ticket
How to Apply

Send me the following 4 things — applications missing any item will be ignored:

A 2-minute Loom video introducing yourself, walking through your strongest AI / Python project
Your GitHub profile link
Your monthly salary ask (USD)
Answer this question in writing: "What's the difference between a tool call and a structured output in the Claude API, and when would you use each?"
Will be reviewing top canadidates through a project test.

We're hiring this ASAP