Case Study
Rideshare Collision Network
A multi-platform system that connected rideshare drivers with collision repair shops. Built entirely by hand, before AI-assisted development existed. Discontinued when COVID-19 collapsed the rideshare market.
Integrated Platforms
Database Tables
Hand-Written PHP Files
In leads from one shop in under 3 months
What I Built
Rideshare drivers get into fender benders constantly. RCN was a platform that connected them with collision repair shops. Drivers submitted damage photos and details, and the system matched them with nearby shops that could handle the work.
The platform included three separate applications: a consumer-facing site for drivers, custom websites for each body shop (with embeddable widgets for shops that already had sites), and a management dashboard where shop owners tracked leads from first contact through repair completion.
At the first shop that used it, it generated over $20K in leads in just a few months before COVID-19 killed the rideshare market and we discontinued it. Every line of code was written by hand. This was before AI-assisted development existed, so it was just documentation, Stack Overflow, and persistence.
Three Platforms, One System
Driver Portal
Rideshare drivers submitted damage photos, vehicle details, and location. The system matched them with qualified shops and facilitated communication through repair completion.
Shop Websites
Each body shop got a custom website (or embeddable widget) connected to the central lead engine. Leads from their own site and from the RCN marketplace all landed in one place.
Management Dashboard
Shop managers tracked every lead from inquiry to closure. Owners monitored response times, conversion rates, and revenue. Role-based permissions kept everyone in their lane.
How a Lead Moved Through the System
Driver Submits Request
Photos, vehicle info, and location entered through any of the three entry points.
System Matches Shops
Geographic proximity, capacity, and specialization scored to find the best matches.
Shops Receive Lead
Matched shops got notifications with all details. Messaging enabled direct communication.
Track to Closure
Every lead tracked from first contact through deal won or lost, with reasons captured for analytics.
Hard Problems I Solved
No AI assistance, no code generation. Just architecture decisions and manual implementation.
Multi-Tenant Architecture
Three separate applications (consumer, business, admin) needed to share data and business logic while keeping their user experiences distinct.
Intelligent Lead Matching
Drivers needed to be matched with the right shops based on location, capacity, specialization, and past performance. Not just nearest. Best fit.
Granular Permissions
Four user types (drivers, managers, owners, admins) each needed different access levels. Owners could see everything. Managers only their assigned leads.
Real-Time Analytics
Shop owners needed dashboards showing conversion rates, response times, and revenue. All computed from 50+ tables with complex joins.
Technology Stack
Backend
Frontend
Database
Why This Matters Today
I built this entire system by hand, before AI tools existed. That foundation of real architecture experience is what makes AI-assisted development so much more effective now. I know what good code looks like because I wrote thousands of lines of it manually. Now I build faster, but with the same attention to getting it right.