Backend / Automation Engineer (AI Systems)

Remote · Contract (Part-Time to Full-Time) · US Eastern Hours Preferred · $8–$12/hr

We build production-grade AI automation systems for property-management companies and SMBs.

Not demos. Not slideware. Real systems that run unattended in production.

Our projects include:

• AI-powered email triage
• Document and invoice processing
• Multi-system data synchronization
• Human-in-the-loop review apps
• Workflow orchestration and automation
• Voice intake systems
• Browser automation and scraping where APIs don't exist

We move fast, ship end-to-end, and care deeply about reliability.

Why This Role Exists

We're growing faster than we can build.

We need an engineer who can take a scoped brief, own implementation, and deliver production-ready systems with minimal supervision. You won't be handed isolated tickets or busywork — you'll build systems businesses actually depend on.

If you enjoy integrations, debugging ugly APIs, automation reliability, and shipping practical AI systems, this role is for you.

What You'll Build

Current flagship engagement:
A multi-tenant automation platform for a commercial property-management firm that:

• Syncs Rent Manager and Microsoft 365 data into a local datastore
• Triages inbound maintenance emails
• Determines tenant-vs-landlord responsibility using lease data
• Detects utility-cost anomalies
• Generates organized audit invoice packages on demand
• Routes everything through a human review workflow

Upcoming systems include:

• Twilio + ElevenLabs voice intake
• Playwright/Skyvern browser automation
• AppFolio and Rent Manager write-backs
• AI-assisted operations tooling

What You'll Own

• Build scoped client automation systems end-to-end
• Design and harden integration pipelines
• Implement retries, idempotency, dead-letter handling, and monitoring
• Work with third-party APIs and undocumented edge cases
• Build deterministic guardrails around LLM workflows
• Write tests and validation fixtures for business logic
• Debug real production issues quickly and pragmatically

Must-Haves

• Comfortable shipping with:
  • Claude SDK
  • Supabase
  • Vercel
  • Git
• Strong API integration experience:
  • OAuth/token auth
  • Pagination
  • Webhooks
  • Rate limits
  • Retry logic
• Reliability mindset:
  • Idempotency
  • Failure recovery
  • Silent-failure detection
• Solid data modeling instincts:
  • Schemas
  • Upserts
  • Multi-tenant scoping
• Able to work from scoped briefs and deliver independently
• Comfortable with AI-assisted development workflows (vibe coding) while still understanding and defending your implementation decisions

Nice-to-Haves

• Microsoft Graph API
• Playwright / browser automation
• Next.js
• PDF/document processing
• Multi-tenant SaaS experience
• Property-management or fintech domain experience

How We Work

• Async-first, low-meeting environment
• Small PRs and clean reviews
• AI-assisted engineering is normal
• Explicit scope boundaries and ownership
• Fast iteration with high accountability

We care less about credentials and more about whether you can reliably ship working systems.

What Success Looks Like in 90 Days

• You've shipped at least one production client workflow end-to-end
• Your systems run without babysitting
• Your PRs are trusted and well-tested
• You can take a brief and deliver independently

Compensation

• $8–$12/hour initially
• Contract role with path to full-time
• Rate increases as responsibility and ownership grow

How to Apply

We do not screen on resumes.

Instead, complete the short proposal exercise below and submit it. Use any AI tools you want — we care that you understand and can defend your decisions.

Application Exercise (Required)

A US commercial property-management firm uses Rent Manager and Microsoft 365. They want three automations:

1. Triage inbound maintenance emails and determine tenant-vs-landlord responsibility
2. Detect utility-cost anomalies month-over-month
3. Generate organized audit invoice packages on demand

Your proposal should cover:

• Feasibility & technical constraints
• Architecture and core design decisions
• Data model sketch
• How you keep AI outputs trustworthy
• Phasing and risk management
• Reliability strategy
• Estimated effort ranges
• Required client access/dependencies

Optional bonus:
A small working spike (e.g. Rent Manager API auth + record pull).

Clean and thoughtful beats large and unfinished.

Strong submissions may lead directly to a paid trial build.