From the first use case to a system people rely on.
My work sits where AI delivery and enterprise program leadership meet. Both come down to the same habits: deciding what is worth doing, making trade-offs explicit and staying accountable until the system runs.
How I think about delivery
I have used the same line of sight on an AI product and on a 13-country banking program: start with the business reason, and end with a result someone can check.
Diagram: Business leads to Transformation. Transformation splits into People, Product and Technology. The three come together in Delivery, which leads to Outcome.
- BusinessThe reason. The problem, the strategy and what is at stake. I always start by understanding why.
- TransformationThe change the business wants to make: what has to be different, and for whom.
- PeopleWho does the work and who is affected: teams, leaders, skills and ways of working.
- ProductWhat gets built and why it matters to its users: scope, priorities and value.
- TechnologyWhat it runs on: architecture, integrations, data, AI and operations.
- DeliveryWhere People, Product and Technology come together: plan, governance, risks, dependencies, releases and KPIs.
- OutcomeThe result that counts: value realized and tracked with a few KPIs, not features shipped.
If one of the three is missing, delivery stalls: people without product direction drift, a product without technology stays a slide, and technology without people is never adopted.
AI delivery and product
I don't start with the model. I start with the process and the outcome we want to improve.
Find the use cases that pay back.
- Map the current process, the pain point and the measurable outcome.
- Check data availability, integration effort, risk and operating cost.
- Decide where deterministic automation beats an LLM, and say so.
Ship working systems, not demos.
- RAG and CAG architectures, LLM and agentic workflows, integrations with APIs and structured data.
- Hands-on low-code and no-code delivery (Bubble.io) when speed matters.
- Build-versus-buy decisions, for example using a scraping provider instead of building one.
Make outputs something you can rely on.
- Evaluation and validation against reference cases.
- Grounding model output in third-party data to reduce hallucination risk.
- Observability, feedback loops and clear human-in-the-loop boundaries.
Keep it affordable and available.
- Model selection, routing, caching and token-usage control.
- Uptime and operational readiness (99.97% on a production GenAI platform).
- Clear ownership and change control for prompts, models and configuration.
AI maturity
A prototype proves that something can work. Production proves that someone can operate it: reliably, under control and with results you can measure.
Enterprise program and delivery leadership
I've seen programs fail because everyone had a plan, but nobody made sure those plans aligned or had the authority to make the conscious trade-offs when they didn't. I build governance that surfaces conflicts early, makes clear who decides, and still leaves teams free to deliver.
Governance and reporting
- RAID, KPIs, readiness reviews and Steering Committees.
- Reporting by exception: what changed, what needs a decision, and my recommendation.
- Evidence that a delivery is under control, not only a green status.
Multi-vendor and multi-country delivery
- One shared plan and one risk log across 9+ vendors and partners.
- Dependency and sequencing management, change windows and rollback planning.
- Influence without formal authority.
Portfolio, capacity and prioritization
- Explicit trade-offs, and one prioritization mechanism applied to everyone.
- Capacity planning, utilization and recruitment planning.
- Portfolio KPIs and margin tracking.
Building delivery organizations
- Operating models, standards and reusable delivery practices.
- A leadership layer of project managers and delivery leads, recruited and mentored.
- From project delivery to managed services: Build, Change, Run.
Commercial accountability
- P&L ownership and margin management.
- Pre-sales, solution shaping and account growth.
- Executive relationships up to CEO, CTO, COO and Chief Security Officer level.
Regulated environments and operations
- Core banking, payments, AML/KYC, identity and API integrations.
- High availability, autoscaling, disaster recovery and observability.
- 24/7 operational governance across markets.
- One platform for many countries: centralized monitoring, one set of access policies and local compliance on a single cloud subscription.
AI still has to be delivered
AI does not remove delivery problems. It makes them surface faster.
In my own AI product I applied enterprise habits from day one: evaluation, cost control and an uptime target.
Technology I have led or worked in
This is the landscape of the environments I have led and worked in, not a list of tools I claim to specialize in.
Cloud and platform
Integration and data
AI and automation
Operations and delivery
Roles I have held, and can step into
Titles vary from company to company. These are the scopes I have carried and the ones I can take on.
| Role | Status | One line | Evidence |
|---|---|---|---|
| Head of DeliveryDOOK.pro, 2016 to 2019 | Held | Own a delivery organization with its KPIs, P&L and clients. | Work: growing a delivery company |
| Service Delivery Manager and Program ManagerCSHARK, 2019 to 2026 | Held | Lead multi-country, multi-vendor programs from pre-sales to managed services. | Work: banking, organization |
| Director of Product DevelopmentTrans.eu, 2016 | Held | Align product and engineering delivery in a large SaaS organization. | About: timeline |
| AI Product and Technology LeadDealWizard.ai, 2025 to 2026 | Held | Own vision, roadmap and technical delivery of an AI product. | Work: DealWizard.ai |
| Senior Project ManagerObjectivity, 2012 to 2016 | Held | Deliver enterprise programs with scope, budget and risk under control. | About: timeline |
| AI Delivery Lead | Can step into | Take AI initiatives from a qualified use case to production across product, engineering and the business. | Work: DealWizard.ai |
| Technical Program Manager | Can step into | Orchestrate engineering, dependencies, rollout and operational readiness across teams. | Work: banking |
| Principal Program Manager | Can step into | Structure ambiguous, multi-stream programs and keep executives informed. | Work: banking |
| Program and Portfolio Manager | Can step into | Prioritize across concurrent programs with one honest view of health. | Work: organization |
Do you have a leadership remit in mind?
Tell me about the remit and where you are in the process.