I help European organisations navigate the strategic, technical, ethical and regulatory challenges of AI – developing sound strategy, deploying AI safely at scale, and building governance aligned with the EU AI Act.
I work with organisations navigating the technical, strategic, ethical and regulatory complexity of AI – helping leaders make sound decisions about what to build, how to govern it, and what responsibilities arise from its use.
My background combines decades of senior technology leadership – including risk and compliance responsibility at Credit Suisse – with deep hands-on engineering experience. I understand how AI systems actually work, and how they must be governed.
Practical support for shaping AI direction and governance frameworks, including the safeguards required for compliant use of GenAI – human-in-the-loop review and oversight to keep systems reliable and trustworthy.
Clear, independent evaluation of AI systems for regulatory alignment, governance readiness, and technical robustness. Grounded in EU AI Act requirements and real-world operational experience.
Practical translation of ethical commitments and regulatory obligations into workable policy – risk classification, oversight requirements, and the documentation needed to evidence compliant use.
AI Strategy & Delivery Director
I've spent decades as a senior technology leader, combining deep hands-on engineering with responsibility for risk, compliance, and large-scale systems delivery. My roles have included Staff Engineer, CTO, investment-banking COO, and Senior Director of AI. I've led multiple AI product initiatives from design through to deployment.
I partner with executive teams to shape AI strategy, and work hands-on with engineering teams to build the first working prototypes together with the operational safeguards they require. This ensures the organisation can adopt AI in ways that are technically robust, trustworthy, and properly governed.
AI is powerful, but it isn't magic. I build working prototypes that teams can validate early, so that organisations can see how a system behaves, where risks may arise, and what operational safeguards will be needed.
My work is grounded in early risk visibility, close collaboration with domain experts, and the principles of trustworthy AI. I recognise the profound impact GenAI can have, and I am committed to maximising its benefits while minimising its potential for harm.
Assess your organisation's AI readiness in just a few minutes.
This tool offers a clear, structured snapshot of where you stand – across strategy, execution capability, and organisational preparedness.
It is free, anonymous, and needs no sign-up.
Effective strategy = precise scoping.
Strategy for AI integration and governance starts with defining boundaries: what you're building, what you're explicitly not building, who owns decisions, and where oversight occurs.
This includes the governance frameworks themselves: what oversight is required, which safeguards are proportionate, and how accountability functions in practice. Ambiguity creates risk, and clear scope prevents ambiguity.
What organisations need
Clear boundaries enable confident action.
When organisations know what they're building, what they're not building, and what governance exists around those choices, teams can direct their energy to solving problems rather than debating scope. Far from constraining thinking, boundaries create the necessary conditions for it.
I work with organisations to establish this clarity early on: defining regulatory obligations, designing operational frameworks, and ensuring the technical safeguards are in place for trustworthy deployment. The goal is to help teams move forward with confidence, knowing their approach is both technically sound and institutionally defensible.
AI is powerful, and responsibility is the first requirement.
GenAI introduces opaque, non-deterministic components into human-facing systems. These are components whose behaviour cannot be inspected or reliably predicted. This makes structured monitoring and evaluation essential.
Unlike traditional software, generative models cannot be reasoned about through code, inputs, or deterministic rules. Safety guardrails must be designed into GenAI implementations from the beginning; and meaningful oversight requires inspectability, automated evaluation, and human-in-the-loop escalation, after deployment.
Why GenAI requires a new standard of risk management
The more powerful the capability, the more rigorous the safeguards must be.
GenAI risk is real, but it is not unmanageable. The same properties that make these systems unpredictable also enable strong technical support for mitigation: AI-based monitoring of interactions with alerting pathways to human-in-the-loop review.
Most of the safeguards organisations need are well-established patterns in modern AI architecture – and, when designed in, they bring non-deterministic systems back within the boundaries of accountable, inspectable behaviour.
I help organisations implement these safeguards in a way that aligns with their obligations under the EU AI Act. My role is to make risks visible early, recommend proportionate controls, and support teams in building systems that are safe, operationally robust, and trustworthy.
Ethical intent only becomes meaningful when it is written into policy people can actually follow
Most organisations already know the principles they want to stand behind – fairness, transparency, accountability, human oversight. The harder work is turning those commitments into concrete rules, thresholds, and review steps that hold up under regulatory scrutiny and still fit the way teams work day to day.
Where policy work makes the difference
Good policy is enabling, not obstructive – it tells teams what they may do, and within which limits.
Policy written in isolation tends to be either so cautious that delivery is forestalled or so vague that it offers no institutional protection at all. Written alongside the people building and operating the systems, it can become a working reference: clear boundaries, agreed trade-offs, and a shared understanding of where the genuine risks lie.
I help organisations draft, test, and maintain that material – mapping regulatory obligations onto real use cases, framing ethical positions that can withstand challenge, and keeping the policy guidelines current as both the technology and the regulations continue to evolve.
Most people can't form a realistic mental model of GenAI until they try it in context
A simple working prototype often achieves in minutes what hours of conversation cannot: clarity, alignment, and a shared sense of what's possible. It turns abstract thinking into something concrete – a model leaders and domain experts can evaluate, question, and refine together.
Benefits of early hands-on interaction
A good prototype is an instrument of discovery – it aims to engage and reveal.
With each quick, contained iteration, assumptions become clearer, risks surface, and the shape of a workable proposition emerges. This early clarity strengthens every subsequent stage: architecture, data preparation, safety design, governance decisions, and the conversations that determine whether a project should proceed at all.
I support organisations through this process in a steady, structured way – building simple, well-bounded prototypes, refining them with domain experts, and ensuring the insights feed directly into practical next steps. The goal is not to race ahead heedlessly, but to help teams move forward with confidence, guided by the realisations that emerge through the process.
