Selected work
Three levels of evidence, in one place.
This page is not a portfolio. It is a controlled overview of the forms in which applied AI work at this studio can be shown: one internal system open end-to-end, one real client engagement under mutual redaction, and a library of recurring delivery patterns.
// How the evidence is organised
- 01
Internal case
Architecture and governance model openly documented
A system I operate myself. Architecture and governance model are disclosed; system identity withheld. No customer data involved.
- 02
Redacted client case
Real engagement, architecture disclosed, names and numbers redacted
A specific client engagement where the governance and architecture are disclosable in the abstract, while client identity, commercial terms, and specific parameters are kept under mutual agreement.
- 03
Reference engagements
Representative delivery patterns, not named cases
Patterns that recur in applied AI work — written up as shapes of delivery rather than individual customer stories. Figures reflect what the pattern typically produces when operational discipline is applied.
Real cases · tiers 01 and 02
2 cases// 01 · Internal · openly documented
An in-house wiki for production AI systems
A human-governed knowledge system built on Claude Code. Decisions, patterns, and operational knowledge are documented, reviewed, and structured so that AI systems in production can reliably build on them.
// 02 · Client · redacted
Vertical B2B comparison portal
An operator-bias problem solved inside the scoring architecture, not the copy. Evidence-first principles, typed confidence taxonomy, numbered defect log, validation harness.
Tier 03 · reference engagements
Delivery patterns that recur across engagements.
Below the two named cases sits a library of recurring delivery patterns — how specific system shapes are scoped, built, and handed over. These are not individual customer stories. They describe the forms of work that come up repeatedly when the operational discipline described on the About page is applied end-to-end.
// Beyond what is public
More is available, just not on this page.
Named versions of redacted cases, additional implementation detail, and engagements that do not appear here at all are discussed under NDA in scoping conversations. The public page exists so that a serious buyer can decide, before that call, whether the discipline shown here is a fit for the problem they bring.
- —Named version of the redacted client case, under NDA.
- —Additional implementation detail for either case, on request.
- —A scoping conversation about a similar situation in your own system.