I design and build the full data chain: from measurement plans, tracking and source integrations to cloud infrastructure, reliable data models, dashboards, automation and machine learning.
Not a collection of loose integrations, but one manageable system that keeps enabling new data products.
Reliable data does not start in a dashboard or data warehouse. It starts with knowing which data is created at the source, how it is collected, and under which definitions it may be used later.
The pipeline shows how each layer builds on the previous one.
01 — Generate & collect
01
Before data can be stored, it first has to be created or unlocked reliably. I design measurement plans, event taxonomies and first-party dataLayer structures, and connect existing operational sources such as CRM, ERP, e-commerce and advertising platforms.
→Reliable event data and unlocked source data with a documented meaning.
Data is stored centrally in a cloud environment that is reproducible, secured and cost-efficient. Infrastructure, roles, datasets, secrets and deployments are not managed by hand, but captured as versioned code.
→A manageable cloud foundation inside the client's own environment and accounts.
BigQuery dataset architecture
Terraform and Infrastructure-as-Code
IAM and service accounts
Secret Manager
Partitioning and clustering
Cost and retention policies
Cloud Run infrastructure
Controlled ingress and egress where needed
GCPBigQueryTerraformCloud RunIAMSecret Manager
dbt build03
03 — Model & trust
modelssource → staging → marts
testspassed
lineagedocumented
freshnessmonitored
contractsenforced
Raw data only becomes usable once definitions, keys and quality rules are made explicit. I build layered data models in which source data is standardised, tested, documented and turned into reusable business logic.
→Tested and documented data models that serve as a reliable source for multiple applications.
Analytics engineering with dbt
Bronze, Silver and Gold layers where appropriate
Staging and canonical models
KPI definitions
Data quality tests
Lineage and documentation
Incremental processing
Business anomaly detection
Freshness and volume monitoring
dbtSQLBigQueryData QualityTestingObservability
04 — Understand & decide
04
On top of the trusted models sits a single, unambiguous layer for analysis and decision-making. Dashboards are not the foundation here, but one of several interfaces on the same controlled data.
→Insight that traces back to shared definitions and verifiable source data.
KPI and metric governance
Semantic reporting layers
Power BI
Looker Studio
Operational dashboards
Management reporting
Funnel and cohort analyses
Anomaly alerts
Scenario analysis
Power BILooker StudioKPI LayerAnalyticsAlerting
activate_pipeline.py05
05 — Activate & automate
triggerscheduled | event_driven
destinationcrm | ads | api | workflow
modelspredictive
monitoringactive
feedbackenabled
The value of data does not stop at reporting. In the final layer, insights flow back into processes, platforms and predictive applications.
→Data that drives processes, enables predictions and feeds new data back into the system.
The architecture stays recognisable, but the concrete solution differs per organisation. Within each layer I build both common foundations and specialised data products.
Measurement plans, first-party dataLayer and (server-side) web analytics — including custom first-party trackers and collection APIs where needed.
Input
Business goals, customer interactions, processes and measurement questions.
Control
Measurement plan, event taxonomy, consent boundaries, schemas and naming conventions.
Output
Reliable event data and a documented tracking contract.
From dashboards and alerts to reverse ETL, marketing APIs and predictive models on Vertex AI.
Input
Validated data models and operational signals.
Control
Authorisation, model validation, consent, monitoring, retries and feedback loops.
Output
Dashboards, APIs, workflows, marketing activation and predictive models.
Power BILooker StudioReverse ETLMarketing APIsVertex AIMMM
← activation feeds generation
03 // CUSTOM ENGINEERING
Custom engineering for the hard parts.
When standard connectors or dashboards fall short, I design focused solutions that still fit the same manageable cloud architecture.
Custom tracking and collection pipelines when control, privacy, data quality or vendor independence is what matters.
Custom web tracker
FastAPI collection endpoints
Server-side event validation
Schema contracts
Bot and duplicate filtering
First-party storage
Custom work for dynamic data sources, non-standard APIs and processes that cannot be unlocked reliably with standard connectors.
Complex API workflows
Dynamic public data sources
Headless browser extraction where necessary
SKU- and entity-based extraction
Fault-tolerant ingestion
Continuous quality control
Technical solutions for problems where classic reporting is not enough.
Business anomaly detection
Predictive modeling
CLTV and churn
Bayesian marketing models
Workflow automation
Distributed Cloud Run services
Observability across multiple modules
standard
standard patterns where possible
custom
custom engineering where necessary
ownership
ownership by design
04 // SYSTEM OWNERSHIP
No black box. A foundation that stays yours.
The result is not just a dashboard or a working integration. You get a transferable data system that can be managed and extended within your own cloud environment.
Your own cloud foundation
The infrastructure runs inside your accounts, under your access policies, without unnecessary dependence on an external platform.
Version-controlled infrastructure
Cloud resources, configuration and deployments are captured reproducibly as code.
Tested data models
KPIs and business logic are documented, verifiable and reusable across multiple dashboards, workflows and data products.
Operational control
Logging, monitoring, documentation and clear responsibilities keep the system transparent and manageable.
Support in a way that fits your organisation.
Delivery does not have to be the end of the collaboration. The support model is matched to the available knowledge, capacity and desired level of autonomy — from fully managed to full handover.
managed
Fully managed
I take care of day-to-day technical operations, monitoring, maintenance and further development of the platform.
Monitoring and incident follow-up
Dependency and security updates
Pipeline maintenance
Data quality
Cost monitoring
New integrations and data products
assisted
Managed together
I stay available as a technical sparring partner and support the internal team on architecture decisions, incidents and new developments — the organisation runs the system itself.
Periodic architecture reviews
Code and model reviews
Incident support
Onboarding new team members
Advice on extensions
Temporary engineering capacity
transferred
Knowledge transfer
I document the architecture, working processes and design decisions and transfer the knowledge deliberately, so the internal team can run and evolve the system independently.
Technical documentation
Runbooks
Architecture diagrams
Deployment instructions
Pair programming and workshops
Handover of responsibilities and access
Build the data foundation that creates real value for your organisation.
Whether you eventually run the system yourself, evolve it together with me or have it fully managed: the architecture stays transparent, transferable and built within your own environment.