The engineer behind
the routes.

The short version

I'm a Principal Engineer at FarEye, where I architect the real-time brains of logistics — a scalable, event-driven alerting system (route deviation, geofence, ETA, stoppage, overspeeding), automatic trip closure, live device-ping ingestion at scale, delivery slot-booking, and a redesigned multi-point pickup/delivery model for milk-run and less-than-truckload shipping. It spans both land-based GPS tracking of vehicles and ocean tracking of shipping containers — telematics in real time, across sea and road.

I've been building software for over a decade — from Mobiloitte and U2opia to Engineer.ai and now FarEye — and the constant thread is maps that have to make decisions. That's what pulled me deep into open-source routing: Valhalla and OSRM, OpenStreetMap data, costing models, distance/time matrices and map matching.

And I don't just prototype them. I've taken routing engines to production the hard way — real traffic, real map refreshes, real uptime — running on Kubernetes with autoscaling, health checks and zero-downtime rollouts. That production scar tissue is exactly what I bring to consulting clients.

What I'm actually good at

• Building & productionizing routing engines end-to-end — Valhalla & OSRM on OSM data
• Custom costing models — trucks, restrictions, time-based access, last-mile rules
• Fast N×N distance/time matrices that feed VRP and dispatch
• Map matching, isochrones, snapping and elevation
• GPS ping ingestion & processing at scale — real-time state (moving / idle / stopped)
• Land-based vehicle tracking & ocean container/vessel tracking
• Scalable, event-driven alerting systems (real-time, high-throughput)
• Delivery slot-booking & capacity/feasibility engines
• Kubernetes in depth — HPA autoscaling, rolling deploys, resource tuning, Helm
• Ruby on Rails — admin tooling to production services
• OSM & GTFS data pipelines with zero-downtime refreshes
• Migrating teams off paid Google/Mapbox APIs to self-hosted stacks

Why I consult on the side

Most teams that need self-hosted routing hit the same wall: the demos are easy, but production is a maze of data pipelines, costing quirks, scaling traps and silent correctness bugs. I've already walked that maze — many times — so I help other teams skip the year of trial-and-error and ship a routing stack they own and understand. It keeps my hands on interesting problems, and it saves clients eye-watering API bills.

Off the clock

When I'm not thinking about graphs, I'm usually reading about them anyway — cartography, distributed systems, and the strange beauty of OpenStreetMap. I like tools that are fast, boring and dependable, and code that someone else can pick up without a séance.