Backend & APIs
Services, APIs and background processing — designed around the data model and the failure modes, not around a framework tutorial.
Services
Engineering work we are willing to stand behind.
We take on a small number of engagements at a time, and we work on your system the way we work on our own — owning it end to end, in production, with the consequences of every decision landing on us.
01
Engineering leadership
A Director of Engineering seat, on contract, for companies that need the judgement now and the permanent hire later.
- Running engineering
- Delivery process, technical direction, the hiring bar and the standards the team is held to. We take the role rather than advising on it from outside.
- When it fits
- A team that has outgrown its first few engineers, a founder still acting as de facto CTO, or the gap between one head of engineering and the next. Months rather than years.
- What you are left with
- An organisation that runs without us — decisions written down, a delivery process the team owns, and a successor who is set up rather than dropped into it.
02
Product engineering
End-to-end delivery of production systems, from the first commit to the thing running under real load with real users on it.
Web & mobile clients
Web applications and native mobile apps, including store submission and release management for iOS and Android.
Cloud-native delivery
Serverless and event-driven systems on AWS — infrastructure as code, CI/CD, observability and cost control, defined in version control rather than clicked together in a console.
03
Technical consultancy
For teams that already have something running and need an outside opinion that comes with its hands dirty.
- System & code review
- A structured read of an existing system: where the risk is concentrated, what is genuinely urgent, and what is merely untidy. You get a written assessment with a prioritised list, not a verdict.
- Performance & cost
- Finding out where the latency and the cloud bill actually come from, and fixing the handful of causes that account for most of both.
- Technical due diligence
- An independent assessment of a codebase and team ahead of an investment or acquisition — maintainability, key-person risk, licensing and security exposure.
- Team & process
- Why delivery is slow when the engineers are good: branching and release strategy, environment sprawl, test strategy, and the handovers that quietly cost a week each.
04
Systems architecture
Deciding the expensive things early, and leaving the cheap things open so the team can change its mind later without a rewrite.
- Greenfield design
- Service boundaries, data model, storage and messaging choices, and the cloud topology to run them — sized for the load you actually expect, with a clear path to the load you hope for.
- Modernisation & migration
- Moving a legacy system without a big-bang cutover: an incremental sequence where every step ships, every step is reversible, and the business keeps running throughout.
- Integration architecture
- Making separate systems work as one — payments, identity, third-party APIs and internal services — with explicit contracts and sane behaviour when a dependency is down.
05
AI engineering
Two separate things, both real: shipping AI features that survive contact with production, and using AI to deliver conventional software faster than the headcount would suggest.
- LLM features in production
- Retrieval over your own data, tool use and agentic workflows, built with the parts that usually get skipped — evaluation harnesses, regression suites, cost and latency budgets, and a defined answer for what happens when the model is wrong.
- AI-assisted delivery
- We build with AI tooling as a matter of course. The point is not novelty, it is economics: a senior team with that leverage delivers scope that used to need a much larger one, and you pay for the outcome rather than the headcount.
- Reviewing AI-generated code
- Teams are accumulating large volumes of generated code that nobody has read. We audit it for correctness, security and maintainability, and tell you which parts are safe to keep.
Engagements
How we usually start
A conversation
A call about what the system has to do, what exists today, and what is in the way. No charge, no obligation.
A short paid review
One to two weeks, fixed price, producing a written plan with scope, sequence and cost. It stands on its own even if you stop there.
Delivery
Either a fixed-scope build or an ongoing retained team, working in your repositories and your cloud account from day one.