About

There's a gap in this field: senior architects who no longer build, and AI practitioners who've never run anything in production. I work in the space between — 25 years of enterprise systems, hands-on with the newest cloud and AI tooling, turning new technology into systems that actually hold up in production.

25 years of building things that had to work

Most of my career has been spent as an architect on systems where failure wasn't an option — telecom networks, multi-tenant SaaS platforms, big-data pipelines. At Motorola and Nokia, I spent fourteen years on high-availability systems for Tier-1 telecom, including configuration management for 4G LTE networks. At Accenture, I led cloud-native engineering and built a 50-person team to deliver a multi-tenant SaaS platform. At Thoughtworks, I took mission-critical applications through resiliency and cloud-migration work.

The thread through all of it is the same: distributed systems, real production constraints, and the discipline of knowing your design has to hold when real traffic and real data arrive. That discipline is what I value most, and it's what I've carried into everything since.

Retooling for the AI era

In 2023 I made a deliberate choice — step back and go deep on production AI and machine learning, rather than coast on the skills I already had. It also gave me time for my family, which mattered. I treated it as real work: earning Google Cloud's Professional Cloud Architect and Professional Machine Learning Engineer certifications, and building actual systems instead of tutorials.

I also spent time understanding the industry around the tools — how the chip supply chain, the hyperscalers, the newer specialized clouds, and the MLOps ecosystem fit together. Architecture decisions are better when they're grounded in that reality, not just the technology on its own.

How I work

The way I build hasn't changed, even as the tools have. I care about the things that don't show up in a demo — validation, observability, security, and whether a system will still be maintainable in three years. I'd rather ship something smaller that's genuinely production-grade than something flashy that falls over the first time it meets real data.

I think in trade-offs. Every decision costs something, and the job is to be honest about what you're trading and why. I document those calls, question my own assumptions, and I'm happy to be wrong out loud if it gets to a better answer. And I've learned that good engineering judgment is usually plain-spoken — if I can't explain why a design is right in simple terms, I probably don't understand it well enough yet.

Where I am now

These days I build production AI systems on Google Cloud — fraud detection with explainable predictions, enterprise search, agentic architectures — and I write about what I learn along the way. The projects on this site are real: deployed, validated, and documented honestly, including the parts that were hard.

This is my working laboratory. If something here is useful to you, or you'd like to talk about building systems that hold up in production, I'd be glad to hear from you.

Certifications

Professional Cloud Architect

Google Cloud

Professional ML Engineer

Google Cloud