Software,

intelligence,

& the patience

to engineer

them well.

Custom systems, designed and written by hand. We work in long-lived stacks that we can reason about — Java, Spring, TypeScript, Postgres — and we instrument what we ship so that the people who run it can see what it is doing.

We treat tests as the documentation that compiles, and migrations as decisions, not chores.

We deploy machine learning where it pays its rent — classification, extraction, forecasting, agentic workflows — with attention to evaluation, drift, and the cost of being wrong. No magic, no demos that don't survive contact with production.

We are equally comfortable fine-tuning a small model and wiring a frontier model into a careful agent loop.

Infrastructure as a long-term decision, not a checkout. We design systems that scale where they need to and stay quiet where they don't — because operational simplicity outperforms architectural fashion.

We work across AWS, Azure, GCP, and self-hosted environments, with a strong preference for what an engineer on call at 02:00 can actually understand.

An outside perspective from people who still ship. We sit alongside in-house teams to assess architecture, recover stuck programmes, and make the kind of clear written recommendations that survive being forwarded to the executive.

Engagements run from a three-week audit to a quarterly retainer.