A ground-up rebuild of a real-time dispatch system, delivered incrementally with zero downtime and a modern delivery pipeline.
The legacy dispatch monolith could not keep up with seasonal peaks. Deploys took a weekend and often failed. Growth was capped by the software.
We ran a strangler migration: a new event-driven core behind the existing API, cut over route by route. Every increment shipped to production behind flags.
The new platform sustains 3x peak volume at lower infrastructure cost. The team deploys 5–10 times a day with automated rollback.
A thin vertical slice in week one meant every decision after that was validated against real traffic.
Have a system like this?
Start a conversation