Eighteen years making infrastructure behave.

Hired to lead presales. I started by building it.

I was the company's first dedicated presales hire, so I had to build the function before I could run it. The team was sizing complex cloud deals off stale spreadsheet exports. I gave it a discovery-led method and the bill-of-quantities calculators it still runs on. I now lead nine presales architects across the GCC.

Requests come in through an intake platform. A colleague built the first version to my problem statement and design; I built and own the production version, now used across our presales team and by the digital and data presales teams too. I trained the team on it.

The work starts in a workshop with the customer's technical people, and runs through design, sizing and a proof of concept where one earns its place. I am often the one who catches a vendor's sizing that doesn't fit the workload, before it reaches the proposal. That is the part of the job I am known for.

Presales leads into delivery. I handle the technical objections, and when an opportunity changes shape mid-flight I re-architect it. After signature I work with our delivery team, the PMO and the customer's technical stakeholders until the project completes. I don't build or run the platform we hand over.

My most recent win was a sovereign disaster recovery design for a public-sector customer, as lead architect end to end.

Each deal now runs in its own Claude agent, grounded in that deal's documents and dated decisions, with pricing from the product catalogue and Salesforce. My presales skill and Word proposal engine sit in the team's shared skills repository, where I am a maintainer.

Presales leads into delivery.

Where getting the infrastructure wrong actually matters.

My UK public-sector chapter: technical authority on large health, government and education tenders, turning regulatory-and-requirement soup into infrastructure a client could actually defend. I led an NHS foundation-trust engagement end to end, pairing enterprise object storage with a hyperconverged VMware cluster. I designed it, proposed it, presented it, and won it.

On a national pathology-imaging programme I was the supporting architect under a lead, and I will say that plainly rather than claim the whole thing. My part was the exacting work that keeps a clinical platform safe to build on. I fed the sizing and built the bill of quantities with the hardware vendor. Compatibility against three medical-imaging vendors' guidelines was mine to check as well. Knowing the difference between what you led and what you supported is its own kind of credibility.

Designed it, proposed it, presented it, won it.

Ten years with a foot in the rack and a foot in the P&L.

The decade that forged the technical-and-commercial identity. I owned product pricing and portfolio for a set of hosting brands and ran vendor negotiations, while also leading R&D and solution architecture for the IaaS division and standing behind the sales floor as its technical authority.

The signature of those years was that I was hands-on. In 2014 I built the quote for a GPU-accelerated VDI deal, priced it, took delivery of the servers, racked them myself, installed the operating system, and ran a proof of concept partitioning the GPUs across virtual machines until I hit the driver limits of the day. Later I validated high-density GPU nodes for AI and machine-learning workloads: thermal stability, performance baselines, and the datacentre physics of power and cooling that most cloud architects never touch. For five months I also built a six-person deployments team from nothing, hired it, wrote the processes, and did a great deal of the racking and building myself. That is where the remote-hands reality of a datacentre stopped being theory for me.

Racking GPU servers before 'AI infrastructure' was a phrase.