PayPal
hudl
MurexYour infrastructure. Your data.
Your control.
Working with enterprises and semi-governmental organisations on sovereign AI, including private and air-gapped systems.
An AI platform for enterprises and semi-governmental institutions, built to run inside their own infrastructure.
Open models. On-premises or air-gapped. Data and operations stay in-house, so teams can adopt AI while meeting their security and privacy requirements.
My focus is practical control: where models run, what agents can access, and how teams operate and update systems when internet access is restricted.
MADA approaches a room through the principles behind good design: proportion, light, material and the relationship between objects. The aim is a space with coherence and character, shaped through a conversation.
A floor plan is the starting point. Layout, materials and furniture can then be refined in an editable 3D space.
Complex buildings. Clear finances.
Building P&L, complex cost allocations and resident accounts.
Building budgets, P&L and resident accounts for property managers in Lebanon. Complex allocation rules connect shared costs to each resident’s balance.
Product, implementation and business developed together, with AI-assisted engineering.
If you know, you know.
Essays on product, organisations and how people work together.
Read the essays
hudl assist$0 to $2.5M
First-year revenue
Hudl Assist grew from an early MVP into a new business, bringing strategy and product development together with design, quality, operations and support.
Building the software and building the service had to happen together. Revenue grew from zero to $2.5 million in the first year.
PayPalProduct leadership across
a 120-country remit.
Product leadership across a 120-country EMEA remit: PayPal’s first mobile SDK launch, One Touch, onboarding and Braintree integration.
SkypeLed product for Skype, the largest account in the portfolio, with PayPal as its main payments gateway.
Worked with the ECB and regulators across EMEA to bring payment services to market. The remit covered payment security, authentication, KYC/KYB, AML and FATCA, plus preparation for PSD2 and GDPR.
Development, releases and client delivery
for global financial institutions.
Credit SuisseDelivered business-critical software and major releases to global banks, coordinating engineering and client delivery across the US, EMEA and APAC.
Doubled throughput and accelerated defect resolution by improving the release process—from development and validation to client delivery.
Built a product-led organisation from an agile engineering team.
Launched a refinancing business.
Built cross-functional squads spanning product, design, engineering, partnerships and operations. Established a shared growth model and delivery process, alongside the launch of a refinancing business.
CPO & interim CTO.
Helping build Nigeria’s payment rails.
Tripled retention while building a local payments gateway, card issuing and cross-border e-commerce. Customers could top up in naira, spend in hard currencies on digital goods, and shop in Europe with delivery to Nigeria.
Led product and technology across banking integrations, treasury and commercial operations, including business purchasing from China. Strengthened financial reporting and operating discipline; evaluated blockchain for cross-border settlement.
Product mentoring
for Seedcamp founders.
Investment analysis.
Portfolio support.
Product diligence for eight-figure investment rounds with Next47, through Rockwae. Evaluated the product, market and team to test the investment case, alongside portfolio support.
Bringing the practical questions of product work into the MBA classroom.
Product Games participating schools
WORKSPRINTWorkSprint develops product capability through practical courses and bootcamps, complementing the MBA curriculum work.
Essays on product, leadership and the systems behind both.
What does it take for an organisation to own and operate its AI?
How do we build useful AI systems for environments without internet access?
How much freedom should an agent have, and where should people stay in control?
Trying to see things clearly—and make a few things that matter. I’m interested in how we know what we know, in seeing familiar things from first principles, and in giving ideas a visual form. Building is one way I explore those questions, across several projects and interests.
Issue 01
Give agents a feature map they can check. For each feature, record how to open it, what it should do and what evidence proves it worked. The linked guide shows how to check an action that archives several items at once and asks for a video and screenshots of the result.
Central monitoring rules could disadvantage open models. Once people download a model and run it on their own hardware, its publisher cannot monitor or recall every copy. If rules require those controls, companies could lose the option to run these models on their own systems.
These tools give software teams concrete checks for common code risks, with a runner to apply them automatically.
