Data products from first proof to scaled business impact.
I help insurance, banking and payments leaders turn ambitious AI and machine-learning opportunities into accountable products—bridging customer need, technical delivery, risk requirements and commercial value.
About me
Good products happen when business strategy and technical constraints actually talk to each other. In heavily regulated domains like insurance, banking, and payments, building great features isn't enough—they have to be trustworthy, auditable, and maintainable. My role is to sit in the middle: translating commercial needs into clear engineering scopes, while anchoring everything in rigorous documentation and solid processes to build future-proof products.To lead a technical product well, you don't need to write every line of code, but you do need to understand how things work under the hood.
My approach is simple: I aim to understand 80% of each domain in the full data chain. That way, when I talk to data scientists, developers, or risk and legal teams, I can ask the right questions, understand the constraints, and bridge the gap effectively.
In complex, regulated environments, great code relies on great process. I treat documentation and structured roadmaps not as bureaucratic hurdles, but as core product features. Clear documentation ensures that products remain maintainable long after launch, scalable for future teams, and fully transparent for compliance and stakeholders.
Past projects
Project 1: Scaling AI-Driven Fraud Risk Evaluation (Adyen)
The Goal:
Scale a global machine learning fraud system to protect enterprise merchants without adding friction to the payment flow.
Business Highlights
- Scaled the ML system across 8,000 accounts, driving over 15% YoY margin growth and reaching top-tier product status.
- Delivered an upsell campaign that realized over 60% Premium conversion, adding 4 million in yearly margin.
Technical Highlights & Process
- Collaborated with platform teams to prioritize and develop real-time features using in-house Kafka and Flink-based feature stores for sub-400ms scoring pipelines.
- Established clear product requirements regarding latency, availability, failover rates, and operational feedback workflows.
- Introduced Explainability AI (SHAP approach) to demystify model performance, ensuring high-touch enterprise customers and internal teams could audit and trust model decisions.Project 2: Automated Underwriting & Fraud Detection (Allianz Benelux)
The Goal:
Bring data-driven automation to insurance underwriting and claims fraud detection under strict regulatory frameworks.
Business Highlights:
- Built foundational data collection methods that yielded a 10x increase in positive fraud labels, turning sparse data into actionable insights.
- Managed agile delivery squads and cross-functional roadmaps, balancing heavy market demand with realistic engineering capacity.Technical Highlights & Process:
- Leveraged NLP and Azure-hosted AI language models to automate manual lookups across 15+ data sources for MidCorp underwriting.
- Transitioned from unsupervised clustering to supervised models, achieving a 20x improvement in claims fraud detection over previous baselines.
- Navigated complex regulatory environments by aligning closely with Legal, Privacy, and Security requirements, ensuring every automated solution met strict compliance.
Contact Us
To connect for a potential partnership, please connect with us!