A little about me

I lead engineering at Arbitr, formerly Straker. I currently lead a team of more than 15 engineers and QA staff. My work covers platform architecture, delivery, hiring and developing people, and I stay close to the technical decisions.

From translation models to AI agents

I joined Straker in 2013 as a web developer. I later had the opportunity to work on machine translation: preparing bilingual data, experimenting with word alignment, and training statistical translation engines with Moses. That led to neural translation models with TensorFlow and Keras, and more recently to language models and AI agents.

Across those changes, I keep coming back to the same questions. Is the data useful? How are we evaluating the result? Does the system work for the people who rely on it?

Leading the people behind the systems

I’ve built and led distributed teams across New Zealand, the USA and the Philippines. Developing people is a significant part of the job. I give engineers progressively greater ownership of architecture, requirements and delivery, with support as they learn to lead others.

I care about small teams with clear ownership, simple systems, and shared understanding. AI gives us more ways to implement a change. We still need the judgement to decide what to build, understand what changed, and check that it works.

What I share here

This is a place for selected work and practical reflections on AI engineering and leadership. I write about decisions and trade-offs, including the parts that take longer to understand than to implement.