Atlas FreightFull stack
Route planning that saves a delivery fleet hours every day
A full stack planning tool that turns thousands of daily deliveries into optimised routes for dispatchers and drivers.
- Client
- Atlas Freight
- Industry
- Logistics
- Role
- Full stack engineer
- Year
- 2025
- Duration
- 6 months
Built with
- React
- TypeScript
- Python
- PostGIS
- AWS
Scope
- Product engineering
- Algorithms
- Backend
- Mapping
- Kilometres driven for the same deliveries
- -18%
- Nightly planning, down from three hours
- 20 min
- Cities now running on the platform
- 4
The challenge
Atlas Freight's dispatchers planned 4,000 deliveries a day by hand in spreadsheets. Routes were drawn the night before, changed constantly, and drivers lost hours to backtracking across the city.
The approach
I worked alongside the dispatch team for a week before writing any code, to learn the rules that never made it into a spreadsheet.
- A Python optimisation service that respects delivery windows, vehicle sizes and driver breaks, built on PostGIS road data.
- A React planning board where dispatchers drag stops between routes and see the effect on time and distance instantly.
- A lightweight driver view that works on any phone, even with a patchy signal.
The outcome
Planning time fell from three hours a night to twenty minutes, and drivers now cover the same deliveries with 18% fewer kilometres. The tool runs in four cities.
She spent a week with our dispatchers before writing any code. That is why the tool works the way they think, and why everyone actually uses it.