Context
A regional 3PL ran its dispatch operation on a PHP monolith last touched in 2018. New driver workflows were impossible to ship, and a single broken deploy could halt the warehouse for an afternoon. The team needed modernization without freezing the business for months.
What we built
- Strangler-fig migration with read replicas and dual-writes during cutover.
- Next.js dispatch console with offline-tolerant driver views.
- Event bus on EventBridge for status changes and ELD integration.
- CI/CD pipeline replacing manual FTP deploys with one-click rollback.
- Dashboards and SLOs on the new stack from day one.
Outcomes
- Total migration
- 9 wks
- Deploys per week
- 30x
- Lost workdays
- 0
- Cut dispatch latency 41% during peak load.
- Driver app crashes dropped from ~120/day to under 5/day.
- Engineering team ships a new driver feature each week without escalations.