The problem
An employer considering a shuttle program needs concrete answers: who can be served, where riders should meet, how many vehicles are needed, and what the program will cost.
Operational routing adds another layer: time windows, capacity, and on-time service cannot be ignored just to make the optimization objective look better.
What I owned
My SHARE work also includes a proprietary rich vehicle routing solution and a company-wide data warehouse and analytics platform.
I designed and led the commuter analysis platform and built routing optimization. I also led cross-functional engineering spanning backend, data engineering, infrastructure, and optimization, working with enterprise customers and public agencies.
The commuter platform and routing engine are related systems. The funding outcome belongs to the platform story; cost and service metrics belong to the routing work.
The engineering decisions
Respect real-world feasibility. The platform validates inputs and reports unserviceable riders instead of silently removing them. Clusters account for group size, and pickup points are tied to real places.
Right-size the fleet. Routing and fleet sizing form an iterative process. Vehicle capacity and time windows constrain the answer, and plans can be refined as demand changes.
Make the output a customer decision tool. Road-based route visualization, costed fleet plans, and demographic context turn a list of addresses into a program an employer can assess.
Model the service trade-off. The routing engine combines clustering, load balancing, savings heuristics, and ordering to find practical routes, validated against operational data.
A technically optimal route is not enough. Customers need to see who is served, who is not, and the cost and service consequences of the plan.
How success was evaluated
For routing, I evaluated cost alongside fleet miles, driver hours, and on-time performance. Improving cost while making the service unreliable would miss the operational objective.
For the commuter platform, the outcome was a decision-ready fleet plan that could support customer conversations and program feasibility.
What changed
The routing engine reduced cost 40%, fleet miles 22%, and driver hours 18%, while improving on-time performance by 15 percentage points.
The commuter analysis platform became standalone SaaS and helped the company raise $12M. Across my work at SHARE, I served as a technical partner to 100+ enterprise clients and public agencies.
My infrastructure work also resolved 80+ security vulnerabilities and helped achieve SOC 2 certification.
A public engineering summary. Results are scoped to the project and period described; implementation details are summarized at a high level.