Dynamic Capture
Coverage is point-level, not a per-asset flag. Re-driving captured pavement puts the camera on standby instead of filling the drive with duplicates.
AI-powered pavement intelligence helping cities and counties assess roads, prioritize repairs, and plan budgets with confidence.

Introducing
SurfaceConditionObserverUsingTechnology
Drive a route. SCOUT captures geotagged road imagery, tracks coverage live, and rates pavement condition with AI — the native iOS field app I build for StreetIQ, re-architected from the ground up in Swift and SwiftUI.
The shutter fires only where pavement is still owed. Coverage is computed on the phone, so it works with no signal.
From the first turn of the key to a rated road network, every step happens on the phone first.
Pick an assessment. Its road network and map tiles are already on the phone, so the drive works with no signal.
Geotagged imagery fires at a chosen distance interval, dead-reckoned between GPS fixes so spacing holds at highway speed.
Point-level coverage is computed on device. The shutter only fires where pavement is still owed.
Measurements report live during the drive; imagery uploads in the background and survives relaunch.
AI condition ratings land on every asset, with PASER scales that follow the road’s surface type.

StreetIQ is an Indianapolis-based company applying computer vision and machine learning to street-level imagery, giving public works teams an objective way to score roadway conditions, standardize reporting, and communicate progress to stakeholders. SCOUT is where that imagery comes from: the app in the field operator's hands, mounted on the dash, driving the roads.
I joined StreetIQ to replace the original cross-platform field app with a native one, and SCOUT is the result: a ground-up SwiftUI rebuild on a clean, testable, layered Swift Package. I own the architecture, the capture pipeline, the coverage engines, the companion-camera integration, the offline map system, and the Xcode Cloud delivery pipeline behind four environments.
A thin app target on top of an umbrella Swift Package of single-responsibility modules. Dependencies point inward: features depend on the design system and core, data implements core's protocols, and core depends on nothing — so the business logic is fast and deterministic to unit-test without a simulator.
Coverage is point-level, not a per-asset flag. Re-driving captured pavement puts the camera on standby instead of filling the drive with duplicates.
Drive a road and watch it get eaten: orange pellets where capture is still owed, teal dots where measurements exist, completion tracked per road.
Low-latency frames ripped from a continuously running camera, wide or ultra-wide lens with remembered zoom, optional dual-camera.
A 3D driving map that looks down the road, the viewfinder in the corner nearest the camera, and every control stacked at the right thumb.
Route to an asset, an assessment, or a long-pressed pin with spoken instructions. The corner item carries the next maneuver, distance, and ETA.
Pair a GoPro once and it reconnects on its own, shoots alongside SCOUT, pauses when you pause, and rests on pavement already covered.
Every assessment on the device gets its own map tiles over Wi-Fi and keeps them current, refreshed in place when roads or tiles change.
A staged thermal governor, speed-adaptive camera streaming, and two-tier stationary sleep so multi-hour metro drives never hit the critical thermal state.
Background uploads that survive relaunch, pinned to Wi-Fi, auto-retrying transient drops, and surfaced through a Live Activity and Lock Screen state.
SCOUT is built by StreetIQ for its field crews and customers. It is StreetIQ's product, not a Molargik Software app, and is described here only to show the work.