Build Production-Grade Middleware Integration Layer for Multi-Brand Home Services Operator

We are seeking a highly skilled Python/FastAPI developer to build a production-grade middleware integration layer for a multi-brand home services operator. This critical system will seamlessly connect HubSpot CRM, QuickBooks Online (qbo), lmn field erp, and dialpad. The solution must be stateful, ensuring proper enterprise data controls, and will utilize PostgreSQL or Supabase for state management, audit logging, and comprehensive exception tracking. Hosting is anticipated on Heroku or an equivalent platform, with all infrastructure client-owned. This is not a no-code automation project; we require a custom-built service designed for robustness and scalability. Core requirements include robust API integrations with Hubspot, qbo, lmn, and dialpad. The system will feature PostgreSQL mapping tables for cross-system record IDs, a timestamped audit log for every integration write, a retry queue with exponential backoff, and a dead-letter queue for unresolved failures. Idempotency logic will be crucial to prevent duplicate records on retry, alongside specific duplicate prevention mechanisms for contacts, companies, deals, jobs, and invoices. A 'No-Regress Logic' is essential to protect trusted HubSpot fields and stages from being overwritten by lower-confidence upstream data. The exception queue must provide a reviewable surface, allowing failed records to be easily inspected, searchable, and replayable, rather than just logged. The project also includes developing a QBO-to-HubSpot variance bridge to compare accounting revenue against CRM closed-won deal value, Dialpad-to-HubSpot association logic for connecting calls to contacts, companies, and deals, and a comprehensive monitoring dashboard for sync health, exception volume, failed writes, and replay status. Additionally, a lightweight server-side Meta CAPI forwarding endpoint is required. The project architect will provide the technical blueprint, detailed field mapping, source-of-truth rules, and business logic. The ideal developer will be responsible for building clean, maintainable production code from these specifications. The final code must be production-ready, thoroughly documented, and easily maintainable by another developer after handoff.