QRNM · Concept prototype

Qatar Road Network Monitor

One live map for every closure, diversion, and roadwork in Qatar.

All event data is illustrative. This is a concept, not a live service.

The gap

Qatar's road information is scattered. Ashghal announces closures in one place. MOI posts incidents in another. Utilities dig on their own schedules. There is no single view — not for a control room, not for the public.

Some countries run national roadworks platforms: every closure and dig on one map, used by authorities, utilities, and drivers. The Gulf has nothing like it. So I built a concept of one, for Qatar first.

What it does

Live network map. 33 illustrative events across Greater Doha, Lusail, and Al Khor. Closures, diversions, maintenance, utility works, permits, and public events.
One dataset, ten viewpoints. A role switcher shows the same network through different eyes: Ashghal's control room, MOI Traffic, emergency response, public transport, logistics, and a commuter. Each role is one config entry, not a separate build.
Journey impact. Save a home and work location. The app checks today's events against your route and tells you what will hit your commute — before you leave. Live traffic apps tell you about congestion while you drive. This tells you about planned disruption in advance.
Clash detection. Pure geometry: any two events within 500 m with overlapping dates get flagged. A road closure on top of a water-main dig is exactly what a shared platform catches and siloed authorities miss.
Address lookup. Zone → street → building, through Qatar's national addressing system (QNAS).
Bilingual by design. The chrome uses stacked Arabic-over-English labels, the same pattern as Qatari road signs. The AI assistant answers questions in either language — powered by Fanar, Qatar's own LLM built by QCRI.

How it's built

Next.js and React. Leaflet with CARTO basemaps. Turf.js for the spatial maths. Routing runs through a three-tier fallback: live routing API, cached route, straight-line corridor. If everything upstream fails, the app still answers. Deployed on Cloudflare Workers.

The AI layer is deliberately split. Clash detection and reports are deterministic — distance and date maths, no model, no way to hallucinate. The language model handles the one job that needs one: answering free-text questions, in Arabic or English, grounded only in the event data. If Fanar is unreachable, preset questions still get computed answers locally.

The map has to be right

This is a tool about a real city, so it has one hard rule: every coordinate is real. Each of the 33 events sits on a verified location, pulled from live map data and checked by hand. The headline closure sits on the actual Salwa Road, between the actual C-Ring and Expressway interchanges. A Doha local spots a misplaced marker in seconds, and this map is built to survive that look.

The addressing work taught me something along the way. The QNAS zone list returned three entries for zone 5, which looked like duplicate data. It wasn't. In Qatar, one numbered zone contains several named districts — zone 5 holds Fereej Al Asmakh, Al Najada, and Barahat Al Jufairi. The right fix was grouping them, not deleting real places. When data looks wrong, check your assumptions before you "clean" it.

The honest part

The software was never the hard problem. The data is.

Platforms like this work abroad because regulation forces the data to exist — authorities must register street works, and the platform sits on top of that. Qatar has no such mandate. A version of this with real data needs either the authorities as customers, publishing through their own tool, or patient aggregation of what they already announce in public.

That's a business-development problem, not a code problem. This prototype exists to make the conversation concrete: here is what the platform looks like, here is what the commuter gets, here is what the control room gets. The build is done. The interesting question is who wants it.

What I'd do next

Bake real road geometry into routes and events (the prototype uses point markers and corridor approximation).
Full Arabic mode with RTL layout and Arabic map labels.
Scenario datasets: a major incident, a National Day event, severe weather.
A pilot with one authority's real published data.

Built by Shiraz Hussein. Planned with Claude, built with Claude Code, reviewed and tested by hand.