Start Here

Over a hundred technical articles, and the handful that best show what I do. Architecture judgement, AI infrastructure, resilience, and the tools I built and gave away.

Start Here. A hundred articles, one short list, grouped by the kind of problem each one deals with.

There are over a hundred technical articles on this site. Nobody should have to browse that to work out whether any of it is useful. This is the short list, grouped by the kind of problem each one deals with.

If you only read one
The Platform Decision. It is an evaluation framework that deliberately does not come with a recommended answer, which is the opposite of how most vendor-adjacent writing works and the piece I would most want someone to judge me on.

Architecture and the decisions behind it

How to run a platform evaluation without a predetermined answer. Scoring, weighting, and how to stop the exercise becoming a justification for a decision already taken.
Sequencing a migration around a compliance deadline without wrecking the adoption that follows. Four stopping points and the trade-off at each.
The same migration at build level: topology, network, storage, management plane and integrations at every stage. Written for the people doing it rather than the people approving it.
The decisions that have to be made before anything gets built, from watching the same questions surface too late on project after project.
Where architecture meets the commercial conversation. What a finance team will actually challenge and how to have the answer ready.

AI infrastructure, below the marketing layer

Why utilisation numbers mislead and what the hardware is really doing. The opening of a seven-part series.
Prefill is compute-bound, decode is memory-bandwidth-bound, and almost every capacity conversation gets this backwards.
Cost-per-token modelled honestly, including the parts vendors leave out of the comparison.
Treating retrieval as infrastructure rather than as a feature, and what that changes about how you build it.

Security, compliance and resilience

What post-quantum readiness means for an estate you are designing today, rather than for a conference slide.
A critical advisory mapped to a remediation path for every supported stream. The kind of work an operations team needs within hours, not weeks.
A compliance control implemented properly, and what it costs you operationally to run it.

Things I built and gave away

Fifteen questions across five pillars, scored in your browser. No sign-up, nothing transmitted. Built because the alternative is a consultant and three workshops.
All 188 Las Vegas breakouts, categorised and described, each linked to the official recording. With a companion spreadsheet indexing 246 speakers.
The same catalogue read as evidence: what the session mix says about where the product is actually going.

Getting value from what you already own

Entitlements sitting unused, and why that is usually an operating model problem rather than a technical one.
Getting multi-tenancy right early, because retrofitting it is the expensive version.
Licence portability across AVS, GCVE, OCVS and EVS, and what it means for where workloads can credibly go.

If you want the rest, the Architecture, AI and Security sections hold everything, and new posts go out by email as they publish.

If something here is wrong, or you are running one of these designs in production and it did not hold, I would rather know. Contact details are on the About page.