Fleet Procurement Optimization

Managing a fleet of over 15,000 vehicles across multiple regions required constant procurement decisions: when to buy, what to buy, and where to deploy. The fleet included large vans, standard vans, ICE vehicles, and an expanding EV programme. Decision-making was fragmented across spreadsheets, gut feel, and siloed regional teams with no unified view of demand, station constraints, or total cost of ownership.

I built a Python-based optimization model that ingested demand profiles, geographical constraints, station capacity limits, driver availability, maintenance schedules, and EV-ready routes to recommend the optimal vehicle mix for each region. The model provided recommendations on optimal procurement decisions for purchasing large vans vs standard vans, ICE vans vs EV vans, evaluating trade-offs across total cost of ownership including fuel/charging costs, maintenance, depreciation, regulatory incentives, route distances, payload requirements, and charging infrastructure availability.

The model was deployed serverlessly on AWS using Lambda and API Gateway, enabling on-demand execution and scenario planning. Model results were fed into a QuickSight dashboard for real-time tracking, giving leadership visibility into fleet composition, utilization gaps, and procurement recommendations. The cross-functional team of analysts and scientists I led delivered £80M+ in capital expenditure savings in 2024.