The person behind the architecture

Built on
curiosity.
Driven by craft.

I've spent 15+ years at the intersection of engineering and possibility — the place where a well-designed system unlocks something that wasn't achievable before. First that was distributed systems. Then machine learning. Now it's agentic AI at enterprise scale.

What hasn't changed is the approach: understand the problem deeply, design for the humans involved, build for the edge cases, and always leave the system better than you found it.

"The architecture is never the point — the outcome is."

Current focus
Agentic AI at enterprise scale
Based in
Enterprise · Multi-cloud
Background
Engineering + Org strategy
Years in tech
15+
Primary model
Anthropic Claude
Status
● Available
01

The Origin Story

It started with a distributed systems problem that wouldn't behave. Years ago, working on a platform that needed to serve millions of requests per day with near-zero latency, I realised that the interesting part wasn't the technology itself — it was the gap between what the system did and what the humans using it actually needed.

"Good architecture makes the right thing easy and the wrong thing hard. Great architecture makes people not notice the architecture at all."

When machine learning started reshaping what systems could do, I saw the same pattern: the models were impressive, but the real leverage was in how they were deployed, integrated, and governed. How they fit into the organisations and workflows they were meant to serve.

That's what led me to AI/ML architecture — not just building the models or running the infrastructure, but designing the full system: the data pipelines, the serving layer, the human interfaces, the governance frameworks, and the feedback loops that make everything improve over time.

Now with agentic AI, we're at another inflection point. The question isn't whether LLMs are capable — it's whether we can build the scaffolding that lets them act reliably, safely, and usefully at enterprise scale. That's the problem I work on every day.

02

The Journey

2022 — Present
Chief Architect, AI Engineering
Leading enterprise AI strategy across 12+ business units. Designing the agentic AI platform, MCP ecosystem, multi-cloud LLM orchestration, and AI governance frameworks. Executive liaison between engineering, product, and C-suite on all AI/ML initiatives.
2019 — 2022
Principal AI/ML Architect
Architected platform-agnostic ML pipelines across GCP, AWS, and Azure. Established enterprise MLOps practices — CI/CD for model lifecycle, drift detection, automated retraining pipelines. Early adopter of HuggingFace Transformers and PEFT fine-tuning techniques.
2015 — 2019
Senior Data & ML Engineer
Built the data infrastructure that made ML possible — feature stores, real-time pipelines, A/B testing frameworks, and the first production ML models. Learned that the data layer is where most AI projects actually succeed or fail.
2010 — 2015
Distributed Systems Engineer
High-throughput distributed systems, microservices, event-driven architectures. The foundation — understanding how systems behave under load, fail gracefully, and recover. Every AI system I build today still reflects lessons from this period.
03

Architecture Principles

Principle 01
Platform-agnostic by design
No vendor lock-in at the architecture layer. Every system I build can swap cloud providers, LLM vendors, or serving frameworks without a rewrite. The business logic should outlast any single platform's feature roadmap.
Principle 02
Observability is not optional
A system you can't observe is a system you can't trust. Full trace, cost attribution, quality metrics, and safety monitoring ship on day one — not as an afterthought. You can't improve what you can't measure.
Principle 03
Design for humans first
The most technically correct solution that nobody uses has failed. Architecture decisions are ultimately about the humans in the loop — the engineers who maintain it, the operators who run it, the people whose work it changes.
Principle 04
Failure is a design input
Every agentic system will fail in production in ways you didn't predict. The question is whether the failure is graceful, observable, and recoverable. Fault tolerance, circuit breakers, human escalation paths — these are features, not contingencies.
Principle 05
Start boring, earn complexity
The best architecture for month one is almost never the best architecture for year three. Start with the simplest thing that works, instrument it thoroughly, and let real usage patterns tell you where complexity is actually needed.
Principle 06
Governance enables velocity
Teams with clear AI governance frameworks ship faster, not slower. When the guardrails are well-designed, engineers spend less time in approval cycles and more time building. Governance done right is a competitive advantage.
04

Outside the Architecture

📖
Reading & Research
Philosophy of technology, systems thinking, organisational design. The ideas that shape how I think about building things — Donella Meadows on systems, Clayton Christensen on disruption, and everything Anthropic publishes on AI safety.
🎵
Music & Composition
A lifelong connection to music — the discipline of composition maps surprisingly well onto software architecture. Structure and improvisation, constraints and creativity, the gap between the score and the performance.
🌍
Teaching & Mentorship
Running workshops on agentic AI, MCP patterns, and responsible ML for engineering teams. The fastest way to deepen your own understanding is to teach it — and the most rewarding thing in the work.
Let's build
something remarkable.

Whether you're designing an AI platform from scratch, scaling an existing system, or navigating the organisational complexity of enterprise AI adoption — I'd like to hear about the problem.