Back to Projects
Field Operations Platform
Rock County Sealcoat
2026

Rock County Sealcoat Operations Platform

End-to-end field operations platform for a sealcoating contractor (Janesville, WI), covering job intake through live customer tracking. The system syncs jobs from Bitumio CRM, extracts scheduling constraints with OpenAI, runs a weather-aware auto-scheduling engine, orchestrates SMS/email customer lifecycle messaging, tracks crew trucks via Linxup GPS with traffic-aware ETAs, and exposes both an admin ops console and a tokenized public tracking portal—bridging office planning and on-the-ground execution in one stack.

Key Features

  • Admin authentication with JWT access/refresh, bcrypt password hashing, and route-protected dashboard
  • Bitumio CRM sync (GraphQL) with Redis token caching and completion write-back queues
  • AI constraint extraction from free-text job notes (date constraints, deposits) via OpenAI
  • Job queue management with filters, pagination, deposits, force-schedule, splits, coat overrides, and constraint review
  • Multi-step weather-aware scheduling engine (capacity, school districts, route optimization, chase mode, approve/lock)
  • Special-job workflows: striping follow-ups, warranty repairs, subcontracted paving/patching
  • Customer confirmation tracking with follow-ups, escalation, manual confirm, and decline backfill
  • Lifecycle messaging (SMS/email): confirmations, morning dispatch, day-before notices, lot reopen, review requests
  • Inbound Twilio SMS and SendGrid email webhooks with signature verification and confirmation parsing
  • Message Center for inbound/outbound history, unread counts, and status tracking
  • Linxup GPS pipeline: webhooks, Pull API staleness fallback, geofencing, truck state machine, job completion
  • Live admin fleet map and public customer tracking portal with Socket.IO and Google Maps
  • Traffic-aware ETA pipeline (Google Routes API) with Redis caching, circuit breaker, and Haversine cost guardrails
  • Retell AI phone agent integration: call logs, transcripts, audio playback, opportunity sync to Bitumio
  • Weather forecasts (Open-Meteo), auto day exclusion, and weather-cancel with reschedule flows
  • Analytics dashboard, audit log, app settings, subcontractor management, and schedule PDF export

Technical Architecture

  • Modular NestJS monolith with feature modules (auth, Bitumio, scheduling, messaging, GPS, tracking, phone-agent, weather, routing)
  • Service-layer domain logic with event-driven fan-out (@nestjs/event-emitter)
  • BullMQ background workers for messaging, CRM sync/write-back, and webhook processing
  • REST API (/api/v1) + dual Socket.IO namespaces (/gps, /admin/gps)
  • Frontend: component-based App Router UI, domain React Query hooks, dual Axios clients (admin JWT vs public portal), React Context auth
  • GPS truck state machine → job completion → messaging + Bitumio write-back pipeline
  • Composable scheduling pipeline (day selection → mode → capacity → school → route → time windows)

Highlights

This project demonstrates full-stack product engineering for a real field-service business: syncing external CRM data, applying AI to unstructured notes, optimizing daily schedules under weather and capacity constraints, coordinating customers over SMS/email, and closing the loop with live GPS, geofencing, and traffic-aware ETAs.
The architecture combines a modular NestJS backend, Redis/BullMQ async processing, event-driven workflows, and a Next.js ops console plus public tracking portal—backed by substantial automated test coverage around the highest-risk scheduling and messaging domains.