Five ways in, in order.

01 The engagements

1. i

   Review the engineers

   You inherited a team and you don't know what you've got. Or the board is asking whether you need to hire, and your honest answer is a shrug.

   I assess your engineers from their artefacts: pull requests, diffs, review threads, commit history. No interviews, no 1:1s, no tests. Read-only access is the whole ask, so it costs your team zero hours.

   You get a calibration document per engineer, and a summary with a specialties matrix that answers the real question: do I need to hire? Last time the answer was no. The problem was structure, not headcount.

   2–3 weeks. Fixed fee, scoped in a free 30-minute call.

   Ask about an engineer review →
2. ii

   Build the execution plan

   You know the system needs work. What you don't have is a defensible order: one you can take to a board, hand to engineers on Monday, and still believe in three months later.

   I write a sequenced programme, sized in sprints, where every story can be undone within a working day and carries a written safety clause. It lives as markdown in your repo, imports into your tracker, and renders to PDF for the board. It also says what not to touch.

   The last one ran to 14 sprints and 124 stories. It absorbed a scope cut and two staff departures as dated addenda, and was never rewritten.

   4–6 weeks. Fixed fee, scoped in a free 30-minute call. Usually follows the engineer review.

   Ask about an execution plan →
3. iii

   Clear the tech debt

   Every change is a coin flip. The team has stopped proposing refactors because the last one caused an incident.

   The safety net comes first, then the work. A per-file coverage ratchet stops the bleeding. Frozen contract tests make the refactor safe. A migration drill (break something on purpose, watch the tests catch it) proves the net actually holds.

   Last time, the primary backend went from 0 tests to 402 in ten weeks. The god objects grew anyway, and the progress report led with that.

   A fixed fee per two-week sprint, or a monthly retainer. Scoped off the execution plan, never cold.

   Ask about clearing tech debt →
4. iv

   Install the engineering operating system

   You've cleared the debt, or you're clearing it. The question now is whether it comes back the moment I leave.

   This isn't scrum training. It's the things that make engineering legible: a measured baseline with targets committed before the work, one word per concept, a PR format harvested from your best practitioner and enforced in CI, a board that never claims a state that isn't real, and onboarding built as a product.

   It's process and documentation that outlive me. It isn't a training course, and I won't sell it as one.

   A fixed-fee install, or a monthly retainer of 2–4 days. Scoped in a free 30-minute call.

   Ask about the operating system →
5. v

   Fix the architecture

   Your pipeline was built by someone who's gone. Nobody has audited the identity model, and you can't say for sure what your deploy credentials can reach.

   A security and CI/CD audit in two independent stages. The second stage's job is to adjudicate the first. Every finding is a work order: evidence to the file and line, a runnable test, and a rollback. Remediation is quoted separately, off the findings, so the audit stays independent.

   Last time, the second stage's biggest find was the first stage's false clean bill of health. Already have a security report? I'll audit the audit.

   2–3 weeks. Fixed fee, scoped in a free 30-minute call.

   Ask about an architecture audit →