AI-Optimized Logistics Route Planning
Route optimization for cost-effective and faster deliveries.
Key Details
| Challenge | Dispatchers planned routes in spreadsheets; fuel and late deliveries piled up. |
|---|---|
| Solution | An optimizer that respects windows, capacity and live traffic, with a planner UI. |
| Technologies | OR-Tools, Python, Maps APIs, PostgreSQL |
Technologies used
Python
PostgreSQL
Docker
Google Cloud
Redis
Client background
A regional logistics operator planned routes in spreadsheets. Time windows and vehicle capacity were approximated by experience; fuel spend and late deliveries climbed as stop density grew.
Key challenges
- Spreadsheets could not replan when traffic or cancellations hit midday.
- Capacity and time-window constraints were informal and often violated.
- Empty miles and overlapping zones went unnoticed until month-end.
- Dispatchers lacked a UI to compare scenarios before committing drivers.
What we built
- OR-Tools based optimizer respecting windows, capacity and service times.
- Maps API integration for travel times with traffic-aware updates.
- Planner UI for comparing scenarios and locking routes to drivers.
- PostgreSQL operational store with audit of plan changes.
Project team: 5 engineers across AI/ML, backend and domain specialists — delivery over 14 weeks.
How we delivered
01
Capture
Digitized historical stops, constraints and failure modes.
02
Optimize
Tuned objective for cost vs. on-time with dispatcher feedback.
03
UI
Shipped a planner console that fits the morning dispatch ritual.
04
Operate
Added midday replan hooks and exception alerts.
Business impact
LowerCost per stop
FasterOn-time rate
FewerEmpty miles