Last-Mile Logistics Transformation

FastTrack Logistics was losing margin on last-mile operations. Urban congestion and static sorting logic meant drivers got route sheets that did not match actual road conditions. Same-day delivery promises were missed 23% of the time, overtime costs were rising 18% year-on-year, and customer NPS had hit a five-year low.

1. 1

   Unified order-management, traffic, and driver telemetry data into a Snowflake cloud warehouse using dbt-transformed pipelines updated every 5 minutes.

2. 2

   Applied Google OR-Tools (vehicle routing problem solver) constrained by live traffic density, driver hours-of-service rules, and vehicle capacity.

3. 3

   Built a predictive order-density model using XGBoost on 3 years of order history, enabling warehouses to pre-position high-velocity SKUs.

4. 4

   Developed a React Native driver app that re-sequences stops in real time as traffic incidents occur, with one-tap proof-of-delivery.

5. 5

   Created an operations control tower in Tableau streaming live driver positions, exception alerts, and KPI dashboards.

We replaced FastTrack's static routing tool with a live optimisation engine that treats every route as a dynamic problem. It recalculates stop sequences as traffic, order cancellations, and customer availability signals change throughout the day. A warehouse slotting model ensures the highest-probability items are positioned at the nearest pick-face by 05:00 each morning, cutting sort time per parcel by 40%.