Applied AI engineering, grounded in 12 years of shipping software

Background

I've spent over twelve years building software — from early-stage startups finding their technical footing to platforms serving users at scale. For the last few years that work has converged on one thing: applied AI. Large language models changed what's possible, and I've been heads-down ever since, shipping the agents, retrieval systems, and AI features that businesses actually run on.

What I bring isn't just prompt-craft. It's the engineering discipline to take a promising prototype and make it dependable — accurate, observable, secure, and affordable to operate. The gap between an impressive demo and a production system is enormous, and closing it is exactly where my background in systems and security engineering pays off.

What I do

LLM Applications & Agents

Designing and building production assistants and autonomous agents on Claude and OpenAI — tool use, multi-step workflows, and structured outputs you can rely on. I focus on the orchestration, guardrails, and UX that make AI features genuinely useful rather than merely impressive.

RAG, Evals & Data

Building retrieval pipelines that ground models in your data, and the evaluation harnesses that keep quality from silently regressing. Vector search, chunking and retrieval strategy, eval suites, tracing and observability — the measurement layer that turns "it seems to work" into "we know it works."

Production & Security

The unglamorous engineering that makes AI products dependable: robust TypeScript and Python codebases, cloud deployment and CI/CD, cost and latency optimisation, and secure-by-design architecture. My background in threat modelling and security (SOC2/ISO 27001) means AI systems ship without becoming a liability.

Systems & Scale

Underpinning all of it is more than a decade of distributed-systems and SaaS architecture experience — multi-tenant platforms, infrastructure automation, and the resilience patterns that let a product grow from its first users to its thousandth without falling over.

Philosophy

I believe in pragmatic engineering — decisions driven by real constraints and measurable outcomes, not hype. With AI especially, the temptation is to chase the flashiest capability; the discipline is to ship the thing that reliably solves the problem, can be evaluated, and won't surprise you in production.

Every technical decision has business implications, and every business decision has technical consequences. My job is to navigate that intersection so the AI you ship earns its keep.

Let's connect

If you're building an AI product, exploring where LLMs fit in your business, or need someone to take a prototype the last mile to production, I'd love to hear from you — wherever you are in New Zealand or Australia.

Drop me a line and let's build something that works.