Jun 15, 2025 · logistics · python · optimization

Cross-Border Fleet Optimizer actually starts working

First real systems project: turning a dispatch whiteboard problem into a Python tool that plans better routes than I would by hand.

Context. Routing is combinatorial; humans are forced to approximate under stress. I wanted cost-aware plans.

What broke. OR-Tools wiring mistakes + fake straight-line distances made routes technically valid but operationally wrong.

Takeaway. Real road times (OSRM) + cost-based objective is what makes it believable.

When it became real

Hooked up OSRM for real travel times; first time it chose longer km but cheaper total cost was the turning point.