AI Automation Engineer (n8n, AI Agents, Claude Code, Python, Github)

• Build and maintain automations in n8n across Pipedrive, Google Workspace, Connecteam and other APIs
• Build AI agents and LLM workflows for sales, quoting, Finance, HR, reporting and operations
• Use Claude Code daily to write, edit and debug Python and JavaScript
• Make every automation safe to run unattended
• Build everything so the founder can understand and change it himself using Claude Code

Built for handover (important) I use Claude Code myself and need to be able to make changes without you. That means:

• All code and workflow exports live in GitHub repos owned by the company
• A clear folder structure, readable code, and CLAUDE.md / README files explaining how each system works and how to change it safely
• A short runbook for every workflow: what it does, what it touches, how to test it, and how to switch it off
• No hidden logic, personal accounts, or knowledge that only lives in your head

Guardrails you’ll be expected to build in

• Access: least-privilege API keys, secrets kept in credential stores (never in code), all accounts owned by the company

• AI safety:Human approval before AI sends anything to customers or touches quotes or prices Structured outputs with schema validation Protection against prompt injection Limits on which tools each agent can use

• Reliability:A central error workflow with alerts Heartbeat checks for silent failures Idempotency (no duplicate deals, tasks or invoices) Retries with backoff, timeouts, and respect for rate limits

• Change control: test before production, version control, backups, and a change log
• Privacy: careful handling of customer data sent to AI providers, and clear data retention for recordings and transcripts
• Cost: spend caps and alerts on LLM APIs and automation platforms
• Audit trail: logs showing what each automation and AI step actually did

• Deep, hands-on n8 experience, with real workflows running in production
• Comfortable in Python and JavaScript. You don’t need a CS degree, but you must be able to read and understand code, not just paste it.
• Has built AI agents and MCP servers
• Daily use of Claude Code or a similar AI coding tool
• Can spot when AI output is wrong, and verifies it before shipping rather than trusting it blindly
• Solid with REST APIs, webhooks, JSON and Git
• Documents clearly and writes for a non-engineer reader
• Strong troubleshooter who works independently
• Excellent written English

Nice to have

• Pipedrive, Google Apps Script or AppSheet experience

US$25K - 40K, 2. Paid test project: an n8n + LLM workflow with a custom code step, guardrails, and a README/CLAUDE.md so I can modify it myself

1. Walkthrough with the founder, where you explain your build and your code
2. Offer

When applying, include 2-3 examples of automations you’ve built. For each one, tell us:

• what it does,
• what you coded yourself,
• one way you made it safe to run, and
• how someone else could maintain it.

Safe Footpaths is an Australian footpath inspection and remediation business operating in Melbourne, Sydney and Brisbane. We’re building an AI-powered operations stack and need one hands-on builder to own it.