Data & AI solutions leadership
Connecting business objectives, data and AI strategy, architecture and delivery so teams can move with confidence.
- Technical direction
- AI-enabled solutions
- Scalable delivery
Data Solutions Lead · KPMG Luxembourg
Engineering depth. Architectural thinking. Data leadership.
I connect business ambition with dependable data and AI solutions—bringing together two decades of software engineering, team leadership and architecture experience.
20+Years in technology
4Progressive roles at KPMG
360°Engineering to data strategy
About
I work where technology, people and business meet.
I’m Frank, a technology leader with a hands-on engineering foundation and a career shaped by increasingly complex challenges.
At KPMG Luxembourg, I progressed from Senior Software Engineer and Assistant Manager to Software Engineer Supervisor, then into solution architecture—taking care of the data architecture and the systems that power it. Today, as Data Solutions Lead and Data Architect, I bring those perspectives together to help teams build data and AI solutions that are useful, resilient and built to evolve.
My experience includes AI solution development and the integration of AI capabilities into modern data architectures—connecting models and intelligent applications with the governed, reliable data foundations they need.
I value clarity over complexity, thoughtful collaboration over silos, and pragmatic decisions that create lasting value.
See the full progressionWhat I focus on
Turning data complexity into clear direction.
My work spans the path from strategic intent to the systems, practices and teams needed to deliver it.
Intelligent architecture that lasts
Combining AI capabilities with governed data foundations, designed for maintainability, integration and change.
- Data & AI architecture
- Solution design
- Technology foundations
Engineering leadership
Helping engineers do their best work through context, sound practices, clear expectations and trust.
- Team development
- Engineering quality
- Pragmatic execution
Experience
A career built by moving closer to the bigger picture.
From building software to shaping the architecture and leadership around enterprise data.
Oct 2025–Present
KPMG Luxembourg
Data Solutions Lead · Data Architect
Leading the development and evolution of enterprise data solutions with Microsoft Fabric and Microsoft Purview, connecting data architecture, governance, engineering and delivery across teams and stakeholders.
Oct 2024–Sep 2025
KPMG Luxembourg
Solutions Architect
Helped shape and govern the data architecture, cared for the systems and technology foundations that power it, and translated requirements into sustainable solutions.
Oct 2020–Sep 2024
KPMG Luxembourg
Software Engineer Supervisor
Supervised engineering delivery, supported professional growth and promoted sound practices while bridging hands-on technology with team and stakeholder needs.
Oct 2017–Sep 2020
KPMG Luxembourg
Senior Software Engineer · Assistant Manager
Designed, developed and maintained business software, building the technical and domain foundation for later leadership and architecture responsibilities.
2006–2017
ARISA Assurances S.A.
IT Manager · Software Development Manager · Software Developer
Progressed from software development into engineering and IT leadership, managing people, systems, delivery and the day-to-day technology needs of the business.
View education and earlier experience
Education
Wirtschaftsinformatiker (FH)
University of Applied Sciences, Trier · 2001–2005
Specialization in databases and web development.
Diploma thesis · 2005
Husky Injection Molding Systems
Analysis, development and implementation of a preferred-parts management tool based on customer standards.
Early experience
Technology foundations
Database, document management and simulation-system work across Husky, Mutterhaus Trier and Fem-tech.
Capabilities
A broad view, grounded in hands-on experience.
I’ve worked across the technology lifecycle. These are the areas I bring together in my current work—not arbitrary percentage scores, but practical experience.
01
Leadership & delivery
Team development · Stakeholder alignment · Technical direction · Agile delivery
02
Data, AI & architecture
Microsoft Fabric · Microsoft Purview · Data architecture · AI solution architecture · Enterprise data platforms · Data governance
03
Software & AI engineering
AI application development · C# · .NET · TypeScript · Web applications · Relational databases
04
Communication
German · English · French · Translating between business and technology
Reference architecture
A modern data & AI platform, layer by layer.
A generic, vendor-neutral blueprint of how I think about data platforms. It doesn’t describe any employer’s or client’s environment. Select a block to see what it does and the design choice behind it.
Across every layer
Sources
Ingestion
Lakehouse
Serving
Consumption
How to read it
Data moves left to right, from where it’s created to where it’s used.
Raw data lands untouched, gets cleaned and conformed, then curated into business-ready models with one shared set of definitions. Reports, AI and other applications all build on that same foundation.
Governance, security and operations aren’t a final step. They run across every layer, from the first load to the last answer an AI assistant gives.
Sources
Operational sources
What it does. ERP, CRM and SaaS applications, databases, files and event feeds: where data is created, each with its own model, quality and rhythm.
Design choice. Leave source systems alone. Read through APIs, change data capture or exports rather than querying production directly, and agree data contracts with the teams that own them.
Ingestion
Batch ingestion
What it does. Loads data on a schedule (full extracts, incremental loads, change data capture or API pulls) into the platform’s raw zone.
Design choice. Make loads incremental and idempotent, so a failed run can simply be repeated. Land data as it is first and transform later.
Ingestion
Streaming
What it does. Captures events as they happen through a message broker, for cases that can’t wait for the nightly run: monitoring, alerting, live operations.
Design choice. Stream only where low latency really earns its cost. Most decisions are fine with hourly data, and streaming adds operational weight.
Lakehouse
Bronze · raw
What it does. An append-only copy of source data exactly as it arrived, stored with load metadata such as source, time and batch.
Design choice. Keeping raw history makes every downstream table rebuildable and auditable. Storage is cheap; re-extracting years of data from a source often isn’t possible.
Lakehouse
Silver · cleaned
What it does. Validates, deduplicates and types the data, and conforms it across sources: consistent keys, units, time zones and reference data.
Design choice. Put data-quality checks here, as code, and fail loudly. Quality problems are cheapest to fix close to where they enter.
Lakehouse
Gold · curated
What it does. Business-ready tables modelled around how the organisation works (customers, orders, finance), usually as dimensional models or data products.
Design choice. Model for the questions people actually ask, and give every dataset a named owner. Gold data is a product, not a by-product.
Serving
Semantic layer
What it does. One shared definition of measures and business terms (revenue, active customer, margin) used by every report, notebook and AI assistant.
Design choice. Define each metric once, centrally. Ten dashboards with ten definitions of revenue destroy trust faster than any outage.
Consumption
BI & reporting
What it does. Dashboards, reports and self-service analysis for the people making decisions.
Design choice. Self-service works on top of a good semantic layer, not instead of one. Certify the core reports and let people explore around them.
Consumption
AI & machine learning
What it does. Forecasting and classification models, and AI assistants that answer questions from governed company data and documents.
Design choice. AI is only as good as the data and permissions behind it. Ground answers in curated sources, and apply the same access rules as everywhere else.
Consumption
Data sharing & APIs
What it does. Offers curated data to other applications, partners and teams through APIs, shares or exports.
Design choice. Treat data consumers like customers: versioned interfaces, documented contracts and no surprise breaking changes.
Across every layer
Governance
What it does. A catalog, lineage, classification and ownership: knowing what data exists, where it came from, who is responsible for it and how sensitive it is.
Design choice. Build governance into the platform from day one. Automated lineage and classification scale; spreadsheets of policies don’t.
Across every layer
Security & access
What it does. Identity, role- and attribute-based access, encryption and masking of sensitive fields, enforced from raw data all the way to the AI layer.
Design choice. Grant access by role and data classification, not table by table, and apply the same rules whether a person or an AI agent is asking.
Across every layer
Operations
What it does. Orchestration, monitoring, cost control and CI/CD for pipelines and models: the work that keeps a platform healthy after launch.
Design choice. Treat pipelines as software, with version control, automated tests and deployment pipelines. Alert on stale data, not just failed jobs.
Beyond work
“The best technology work is ultimately human: understand the problem, bring people together, and make the complex feel manageable.”
Outside work, family comes first. I’ve played basketball for more than two decades, and I enjoy books, travel, music and the occasional technical side project.
Connect
Thanks for stopping by.
If you’d like to connect around data, AI, architecture, technology or shared interests, send me a message or find me on LinkedIn and GitHub.