RouteOptim: AI-Powered Last-Mile Logistics & VRP Solver

Last-mile delivery accounts for up to 53% of total logistics costs. The objective of RouteOptim was to engineer an autonomous, zero-cost decision-support system that solves multi-vehicle delivery routing while integrating sustainability metrics (fuel consumption, GHG Protocol Scope 1 CO2 reduction, and operating expenses) into operational dispatching.

Traditional distribution planning often relies on manual dispatching or naive sequential routing. This leads to overlapping vehicle trajectories, capacity violations, excess fuel burn, and unmonitored carbon emissions. Moreover, small and medium logistics operators lack access to lightweight, offline-capable optimization engines without expensive recurring SaaS API subscriptions.

Engineered a modular end-to-end Python optimization architecture:

• Core Solver: Google OR-Tools utilizing Path Cheapest Arc initialization and Guided Local Search (GLS) metaheuristics to escape local minima.
• Distance Engine: In-memory Haversine distance matrix generator with zero external API dependencies.
• Geospatial Dashboard: Interactive Folium/Leaflet route visualization embedded in a responsive Streamlit web application.
• API & Architecture: Decoupled FastAPI REST endpoint for headless solver execution.
• Quality Assurance: Automated Pytest test suite, containerized Docker deployment, and GitHub Actions CI/CD workflows.

Achieved double-digit percentage mileage reduction across multi-country benchmark datasets, provided instant quantification of avoided diesel liters and greenhouse gas emissions, and automated the generation of driver-ready vehicle itineraries with downloadable CSV schedules and interactive HTML map exports.