Services
Consulting engagements
Four ways we typically work with clients. Most engagements begin with the first one, because it is usually unclear which of the others is actually needed until someone has looked properly.
Architecture review
An independent assessment of an existing system: where the structural risk sits, which constraints are real and which are habits, and what the realistic options are. The output is a written analysis with prioritised recommendations and the reasoning behind each, not a slide deck. Typically two to four weeks.
AI and machine learning systems
Design and development of systems with models in them. The hard part is rarely the model — it is the surrounding system: data pipelines, evaluation you can trust, inference that meets a latency budget, and enough observability to tell degradation from noise. We work on that surrounding system, and on the model when that is where the problem is.
Technical due diligence
An assessment of a codebase and engineering organisation ahead of an investment, acquisition or partnership. What has been built, what condition it is in, what it would cost to maintain or extend, and which risks are structural rather than cosmetic. Delivered as a written report suitable for a non-engineering audience.
Embedded architecture work
Longer engagements where we work alongside an existing team — designing a subsystem, leading a migration, or acting as an architectural counterpart to internal engineers. Useful when the knowledge needs to stay with the team afterwards rather than leaving with a consultant.
Engagement
Work is normally scoped and priced per engagement rather than hourly, so the cost is known before it starts. We deliberately take on a small number of clients at a time; if we are not the right fit for a piece of work, we will say so early rather than take it on.