YinMo
AI-Powered Vehicle Damage Estimation & On-Demand Logistics Platform
Enterprise mobile ecosystem connecting vehicle owners, insurance adjusters, repair shops, and towing fleets with sub-50ms reactive damage estimation and real-time GPS fleet tracking.
Executive Product Overview & Scope
YinMo transforms the legacy automotive collision and repair lifecycle into an automated mobile workflow. As the sole Flutter Developer on the mobile client (contributing across 246 commits), I engineered the cross-platform application from scratch. The platform serves two distinct user roles: Vehicle Owners who report damage and request towing, and Service Providers/Towing Operators who manage emergency dispatches, calculate labor and OEM parts costs, generate vector PDF estimates, and navigate directly to accident scenes.
Core Engineering Contributions & Technical Execution
Key Technical Features & Highlights
The Engineering Challenge
The primary architectural hurdle was combining high-frequency GPS coordinate streaming with a compute-intensive 30-variable vehicle damage calculation model. Generating multi-page vector PDF invoices with high-resolution vehicle damage diagrams was causing 1.5–2 second UI freezes on budget Android devices. Additionally, continuous GPS tracking was draining device batteries rapidly during long towing shifts.
The Architectural Solution
Decoupled all vector PDF generation and image compression routines into dedicated Dart compute isolates, completely freeing the main UI thread (0% frame drops). For location tracking, engineered an adaptive GPS polling service that monitors accelerometer movement deltas—polling at 1Hz during vehicular movement and switching to low-power geofence mode when stationary. Stabilized the pricing engine using pure immutable models managed through RxDart reactive pipelines with debounced inputs.
Tangible Results & Measurable Impact
Delivered a 99.8% crash-free application on Google Play and Apple App Store. Reduced PDF generation overhead from a 1.8s UI freeze to <180ms background execution. Achieved sub-50ms pricing responsiveness and supported seamless fleet tracking across thousands of on-road miles.