Senior n8n Automation Architect - Multi-Agent AI Systems

I’m helping TeamGrout find a top candidate to join their team flexible for the role of Senior n8n Automation Architect - Multi-Agent AI Systems.

You'll architect and build a multi-agent AI operating system, transforming business operations and data flow.

Compensation:

USD 1K - 2K/month

Location:

Remote (anywhere)

Mission of TeamGrout:

"To deliver high-quality tile and grout restoration services that restore, protect, and refresh residential and commercial spaces with outstanding results."

What makes you a strong candidate:

• You are proficient in n8n, REST API, CRM tools, Back-end development, AI Agent.
• You have the potential to develop in Laravel.
• English - Conversational
• Spanish - Fully fluent

Responsibilities and more:

We are building a high-performance operating system for a fast-growing home service company based in Toronto (tile, grout, and stone restoration).

Role:

- We are looking for a Senior n8n Automation Engineer who can design, build, and optimize end-to-end automations connecting our CRM, Google ecosystem, and communication channels.

- This is not a basic automation role — we need someone who understands business operations, sales pipelines, and real-time data flow.

Tech Stack (Must Have Experience):

- n8n (advanced workflows, webhooks, API integrations).

- REST APIs (build, consume, debug).

- Google Sheets as database layer.

- Webhooks, cron jobs, queue logic.

- Experience integrating with CRMs (custom or SaaS).

- Understanding of backend systems (Laravel is a big plus).

Skills and Understanding:

- Thinks like a systems architect, not just an automation builder.

- Has experience with home service businesses or service-based operations.

- Understands the lead → booking → job → payment lifecycle.

- Understands dispatching technicians efficiently.

- Understands sales + marketing + operations data flow.

- Can propose better systems (not just execute tasks).

Mission:

- Architect and build a multi-agent AI operating system sitting on top of our existing stack, with four specialized agents reporting into a central orchestrator.

Marketing Agent:

- Monitors ad spend, lead quality by source, offline conversion reporting back to Google Ads, ManyChat flow performance, and proposes budget reallocations.

Sales Agent:

- Owns the lead-to-booking lifecycle.

- Handles inbound across Instagram/Messenger/SMS, qualifies, books estimates, reschedules, and escalates only what humans need to touch.

- Runs setter-level coaching on transcripts.

Operations Agent:

- Manages technician routing, availability, day-of dispatch, materials, and exceptions.

- Integrates with Availability Master 2026 and Marluapp.

Admin Agent:

- Handles invoicing reconciliation, commission calc, KPI rollups, daily and weekly executive briefings.

Orchestrator:

- The orchestrator above them is the COO layer: it watches the four agents, resolves conflicts, surfaces decisions that need a human, and reports to Director of each department (and the owner) in the language and cadence I actually want.

- This is not a chatbot.

- This is a workforce.

What you'll actually do in the first 90 days:

- Audit the existing CRM panels and Sheets architecture (full read access provided day one).

- Define the structure to build and the agents boundaries, memory model, and handoff protocol between agents.

- Build the orchestrator (likely n8n + Claude API + a state store — open to your proposal).

- Migrate the existing classification pipeline into the Sales Agent.

- Wire the Marketing Agent to Google Ads API for offline conversion reporting.

- Ship one fully autonomous loop (inbound lead → booked appointment → closer assigned → ops notified → outcome logged → admin reconciled) before touching the other three.

- Document everything in Spanish and English, because this will be handed to operators, not engineers.

Must-haves:

- Proven multi-agent system experience.

- Not "I read the Anthropic docs."

- Shipped systems with agents that hand off to each other, with memory, with tool use.

- Show me the repo or the workflow.

- Strong with Anthropic API specifically (tool use, computer use is a plus, structured outputs, prompt caching, batch).

- We're a Claude shop.

- n8n at a senior level.

- You know when to use the native nodes vs. an HTTP Request to the API directly.

- You've hit Cloud Starter limits and know the workarounds.

- Google Apps Script proficiency.

- Caja sandbox, ES5 frontend constraints, withSuccessHandler, JSON.stringify boundary rules — you don't need me to explain any of this.

- Google Sheets as a database.

- You know its limits and you know when to escape to a real DB.

- Spanish AND English fluency.

- The team is bilingual.

- The owner thinks in Spanish.

- Bias to ship.

- We move fast.

- Pretty architecture diagrams that take three weeks before any working code do not work here.

Nice-to-haves:

- Experience with JustCall, ManyChat, or any field service / trades CRM.

- Google Ads API experience, especially offline conversion imports.

- Built something with Claude Code or Claude in Chrome before.

- Vector store / RAG experience for institutional memory (our 10,000+ historical client records are sitting in Sheets).

- Comfortable with Marluapp or similar operational tools (you'll be talking to its developer).

Hard nos:

- Agencies subcontracting to juniors.

- I want to talk to the person doing the work.

- "AI consultants" whose portfolio is decks.

- Show working systems.

- Anyone who needs to be told what to do next.

Your potential leader(s):

• Ronald Castillo -