Case Study · Location Intelligence

Finder VTS

Real-time tracking for 15,000+ vehicles with sub-second location updates— built for fleet operators who could not afford map lag or offline gaps.

Mobile Backend Maps

Industry: Logistics / Fleet Management
Duration: Ongoing product partnership
Team: Embedded engineering team
Stack: Flutter · Node.js · PostgreSQL · Google Maps

15,000+ vehicles tracked in production

30% fuel cost reduction reported by operators

Challenge

No live picture of the fleet

Dispatchers managed vehicles through phone calls and spreadsheets. There was no reliable view of where assets were, how drivers behaved, or how much fuel routes actually burned.

The business needed live GPS, geofencing alerts, fuel monitoring, and route insight—without crashing older tablets when thousands of markers hit the map.

Before: dispatcher workflow (artifact placeholder)

Dispatch day (pre-platform)
1. Call driver → ask location
2. Update spreadsheet cell
3. Guess ETA for customer
4. Repeat × fleet size

Failure modes
- stale location data
- no fuel / idle visibility
- no geofence breach alerts

Solution

A map that stays responsive at fleet scale

We built a cross-platform mobile and backend system for live GPS tracking, geofencing, fuel integration, and route optimization—designed so the map remains usable when thousands of vehicles are online.

Product context: Finder GPS Tracker.

Architecture outline (replace with diagram export)

[Mobile apps]
  └─ location ingest + offline queue
        │
[Realtime API / WebSocket]
  └─ fan-out to dispatcher clients
        │
[PostgreSQL]
  └─ trips · geofences · fuel events
        │
[Map render]
  └─ cluster markers below zoom 14

Realtime GPS streaming

Location updates stream over a persistent channel so dispatchers see movement without polling storms that would melt the API under peak load.

Marker clustering

Nearby markers aggregate at lower zoom levels so the map stays interactive when thousands of vehicles are on screen.

Offline-first sync

Field devices queue events locally and reconcile when connectivity returns— critical on routes with patchy coverage.

Configurable geofencing

Operators define virtual boundaries and receive alerts on breach, idle, or route deviation without custom engineering for each customer.

Key decision

Cluster before paint

Rendering every vehicle as an independent marker broke frame budgets on mid-range devices. Clustering at zoom < 14 kept interaction smooth while preserving drill-down accuracy at street level.

ADR excerpt (replace with real review doc)

ADR-012: Map marker clustering
Status: accepted
Decision: Cluster markers below zoom 14
Rationale: maintains interactive map
  under 5,000+ concurrent vehicles
Alternatives considered:
  - server-side tile heatmaps only
  - client pagination of vehicles

15,000+ clients trusting the platform

1,000+ stolen vehicles recovered

30% operational cost reduction target via routing & fuel insight

CI / validation snapshot (replace with real pipeline output)

$ make validate
arch-review ............... ok
map-perf benchmark ........ passed (60fps @ 5k markers)
test-suite ............... passed
deploy/staging ........... ready

Mobile Engineering

Field apps for drivers and dispatchers.

Flutter iOS Android

Backend & Data

Ingest, persistence, and realtime fan-out.

Node.js PostgreSQL WebSocket

Location Intelligence

Maps, geofences, and routing signals.

Google Maps GPS ingest Geofencing

Delivery

Release and health after launch.

CI/CD Monitoring Crash reporting

Start a similar project

Building fleet, field, or location-heavy software? We’ll scope discovery around your constraints—not a generic demo script.

Book a Discovery Call