Expertise
What I bring, and what it rests on
I take on the technology change a company cannot complete on its own, with the team and the plan it has. This page sets out what I actually do and what it has achieved. The situations I am called into are described on the home page.
Four workstreams, run together
An engagement almost always combines four workstreams. Running them one after another does not work.
Diagnose under pressure. Establish what is really happening while the organisation is still trying to explain it, and turn that into a trajectory the leadership can approve and the team can believe in.
Move the product a generation. Not tidy it up: change what the company can sell. Replatforming a legacy suite to the web; putting medical imaging inside a browser tab with React and WebAssembly; defining a cloud-native generation as a business strategy priced against a target cost of ownership; placing an abstraction layer under certified products so they could leave Oracle, cutting customers' database costs by up to 60% and opening a new market segment.
Rebuild the organisation that delivers it. Scrum practised for what it is, with the long horizon and the architecture given back an owner. Teams divided along the lines of the domain, so that several of them can work on one product without becoming one large team in disguise. The articles set out how.
Make the technology credible to those who fund it. I have led the technical demonstration and due diligence behind a majority acquisition by a global investor, taken a regulated product through FDA clearance into the US market, and written the roadmaps leadership teams approve before they commit money. The same honesty that persuades an investor the technology is sound tells a leadership team when it is not.
Software that has to keep working
Most of my career has been spent on long-lived software where mistakes are expensive: platforms that manage a blood supply, settle payments between banks, hold a land registry, or monitor public health in countries with almost no infrastructure. Some of it was regulated; all of it had to stay correct, maintainable and explainable for ten or twenty years.
That shapes how I work. Traceability, a precise vocabulary and an architecture someone owns are not compliance overhead. They are what lets a product survive its third team, its second platform and its first serious audit.
Generative AI, without the hype
I first worked on AI in 1983: speech recognition for patients with severe disabilities, on a single 68000 processor with less than 256 kilobytes of memory. The lesson has not changed. Almost anything can be demonstrated; the real question is which constraint a technology actually removes, at what price, and whether what it produces can be relied upon.
I build with these tools myself, so my advice comes from practice, not from slides. What I see is consistent with the published evidence: generating code has become cheap, while specifying what the software must do and verifying what was produced have become the bottleneck. Organisations that already write down precisely what they mean, with a rigorous domain language and disciplined iteration, are the best placed to benefit. I help leadership teams decide where to invest, where to wait, and how to keep the resulting systems provable.
What it has achieved
| Situation | Outcome |
|---|---|
| A platform nearly four years late, with four clients worth half the revenue at stake | A forty-person taskforce across three countries, a delivery trajectory approved by the leadership, then handover to a permanent CTO |
| A product with no access to the US market | FDA 510(k) clearance; the product went on to lead its US market |
| A company whose product cycle was three years | A product in an entirely new segment shipped in fourteen months |
| Certified products tied to Oracle | Up to 60% lower database cost of ownership, and a new segment of smaller customers |
| A majority acquisition by a global investor | The technical demonstration and due diligence that made it possible |
| R&D organisations of up to 120 people | Led across France, the United States, the United Kingdom, India and the Balkans |
The CV gives the detail of each engagement.
What I am not
Not a project manager, and not an architect who governs from a review board. I am a hands-on CTO who takes architecture decisions, owns them, and answers for the result. I sell no product and represent no vendor.