The layers that now have to be understood together are the ones I have each been accountable for: the systems, the security around them, the data inside them, the products built on top, AI, and the way an organisation actually runs day to day.
I have run networks and infrastructure, information security and risk, data science, delivery and platforms, product, and the executive layer where the question stops being how to build it and becomes which of these deserves the money. The order I did them in matters far less than having done all of them, because it means I can tell when somebody is describing a problem one layer away from where it actually is.
Being comfortable in a board conversation is not the same as understanding what is actually running underneath it. Holding both at once is the whole job, and it is the part that cannot be picked up from one side or the other.
One thing keeps recurring, and I have stopped treating it as coincidence: new technology only produces value once an organisation can operate it, govern it, secure it and connect it to something somebody actually wanted. That has never been more true than with AI, where strategy, data, identity and security stop being separate conversations and become one.
In practice I am fairly direct. I would rather tell you early that a plan will not work than be agreeable about it for three months, and I would rather say a problem does not need me than invent a reason that it does. If you want somebody to validate a decision that has already been made, I am the wrong person to call.
What a system may see, what it may do, which suppliers are involved, and how it is adopted safely.
Networks, infrastructure, platforms, and machines that run themselves.
Information security, risk, privacy, identity.
Data science, modelling, measurement, machine learning.
Product direction, how the pieces fit together, and where the money should go.
What the technology costs to run, what it returns, and how much of that cost turns out to be avoidable rather than inevitable.
Deciding who an AI system is when it acts, what it may see, what it may do on its own, who else handles the information on the way, and whether anyone can reconstruct or undo what it did. Not a feature to be rolled out.
Keeping information safe, keeping it private, controlling who has access to it and deciding how it may be used are now four descriptions of one problem. They are still usually four separate departments.
Technology judgement is often needed for a specific stretch rather than permanently: a reset, a modernisation, a decision that cannot wait. The interesting question is what good judgement looks like when the time is bounded.
Financial, fiduciary, legal and professional advisory settings, where confidentiality, resilience and governance carry more weight than speed.
Listed because people ask, not because any of it is the argument. The work is the argument.
ISO/IEC 27001 Information Security Management Systems and Internal Auditing · Certified Ethical Hacker · EU General Data Protection Regulation Foundation · Data Science and Data Engineering Bootcamp · DevOps Foundation · PRINCE2 Foundation · Certified SAFe Agilist · Certified Scrum Product Owner · Novell Certified Linux Administrator · LPIC-1 Linux Administrator
Away from the work I travel, cook, and read more physics than is strictly useful. The through-line, if there is one, is a long-standing curiosity about how complex systems hold together, and how they come apart.
The work itself is the better introduction.
And if you would rather just describe the situation, that works too.