One engineer, a decade of systems that had to work.

I have spent the last decade building the unglamorous software companies actually run on: ordering platforms, admin consoles, pricing and permission models, and the integrations between systems that were never designed to talk to each other. The work rarely makes a conference talk. It does decide whether a business can take an order on a Monday morning.

What changed recently is the economics. AI-assisted engineering has removed most of the mechanical labour from software — the boilerplate, the migrations, the test scaffolding, the documentation nobody had time for. One engineer with good judgement and the right tooling now delivers what needed a small team three years ago. That is the whole basis of what I offer: team output, single-person accountability, and a price that reflects it.

What has not changed is the part that decides whether a system survives. Someone still has to choose the data model, notice that tenant isolation cannot be left to developer discipline, and say out loud when a requested feature is not worth what it will cost. That judgement is what you are actually buying.

I work with a small number of clients at a time, deliberately. It is the only way to offer a real response time rather than a hopeful one, and it is why every engagement starts with a conversation about fit instead of a proposal.