I'd rather lead something I could still build myself.

The pilot is cheap by design. Production isn't.

Most of the AI conversations I'm in now start after a demo went well. What follows are infrastructure questions and money questions: where it runs, and who pays once real users arrive. I've been doing that work for years under other names.

An AI agent isn't a chatbot. It's an account.

The scope has widened. The technical judgement stayed mine.

I lead nine presales architects across the GCC and co-manage the wider team with the AWS lead. A good part of my week goes on people and on the customer's stakeholders. The rest has a technical centre: workshops, architecture, proofs of concept, the analysis behind a proposal. That part still gets blocked time in the calendar. It is what makes the rest of the job credible.

Azure is my platform. On the AI side I have trialled RAG on Azure AI Foundry and built an Azure OpenAI client for cross-model review. With today's AI tooling the cloud isn't the barrier it used to be, and I am confident delivering on any of them. I'm a doer, not a theorist.

I have taken on more scope over the years and I expect that to continue. What I want to keep hold of is the judgement underneath it. If I am accountable for a decision, I want to have understood it myself rather than taken it on trust. Titles are worth what sits behind them.

Titles are worth what sits behind them.

A vendor's number is a starting point, not an answer.

A recurring thread in my work is sizing: taking a vendor's figures and sense-checking them against real contention ratios and real workloads rather than a spec sheet. It is unglamorous. It is also the thing that stops a design costing someone real money once it ships.

It is a quieter skill than headline architecture, and it is the one clients remember me for. I tend to be the last set of eyes on the detail before a proposal goes out.

Someone has to catch what the sizing missed. That someone is usually me.

I started with a biochemistry degree and no plan to end up here.

I am self-taught on every stack I work with. I moved from an early account-management role into infrastructure because the technical side was the part I kept reaching for, and I learned it by doing the work rather than in a classroom.

Over the years the depth has widened from virtualisation and storage into data, application and security architecture. To keep the hands-on side honest, I run a live production data platform I architected, built and operate on my own, PostgreSQL and PostGIS on Docker, end to end. It is the difference between talking about a system and being on the hook for one.

I design it, I build it, and I run it. All three, or it does not count.