DE

Services

Expertise that turns data into movement.

One senior point of contact from the first question to production, with no hand-offs between consultants. Strategic guidance and practical implementation across the full data lifecycle.

01

Data & AI Strategy

A clear path from fragmented data to dependable AI. I define target architectures, semantic data layers and governance models that connect business priorities with production delivery.

Who it is for: For leadership and technology teams with clear data and AI ambitions but no dependable path from fragmented sources to systems that run in production.

In automotive, this meant shaping the strategy for production machine learning on Azure Databricks for Audi, Porsche and Volkswagen, with governance aligned to GDPR and NIST SP 800-53.

Typical deliverables

Assessment of existing data, systems and the decisions they should supportTarget architecture and semantic data layerPrioritised AI roadmap from first use case to productionModel governance: approval, documentation, monitoring and bias audits

In practice

Automotive AI, from BUS data to decisions
02

Applied Machine Learning

Production-grade models and AI agents for forecasting, classification, behavioral analytics and natural-language data exploration—supported by explainability, evaluation and feedback loops.

Who it is for: For teams that need models working in production—forecasting, classification, behavioral analytics or natural-language data exploration—rather than another prototype.

Past work ranges from retention and churn models that improved marketing return on investment by 28% to an LLM-powered analytics layer for natural-language dataset discovery and PySpark generation.

Typical deliverables

Predictive models with documented evaluationLLM systems and AI agents with review steps and guardrailsExplainability and model reporting for business stakeholdersFeedback loops and A/B tests to measure real impact

In practice

Churn prediction: +28% marketing ROI
03

Data Platforms

Reliable data and ML foundations for high-volume workloads. I build tested pipelines, feature stores and analytical services with Python, Spark, Airflow and Databricks.

Who it is for: For organisations whose data volume or number of sources has outgrown ad-hoc scripts, manual exports and inconsistent reporting.

I have unified CAN, LIN and FlexRay bus data on Azure Databricks, built multi-terabyte Spark platforms and helped migrate ProSiebenSat.1 Media’s analytical warehouse to Apache Spark with a unified KPI reporting layer.

Typical deliverables

Tested data pipelines with Spark, Airflow or DatabricksFeature stores for consistent training and inferenceValidated migration of data, schemas and workloadsMonitoring and observability for data flows

In practice

Warehouse migration to Apache Spark
04

MLOps & Delivery

Operationalise models with reproducible workflows, CI/CD, monitoring and auditable controls. I also mentor teams in clean code, testing and sustainable delivery practices.

Who it is for: For teams that already have models but struggle to deploy, monitor or audit them reliably—or that want to raise their own engineering standards.

At BODYVISER I built secure Airflow and PostgreSQL pipelines for sensitive health data and led technical delivery through the company’s acquisition; in automotive I mentored data scientists in clean code, test-driven development and CI/CD.

Typical deliverables

Reproducible training and deployment workflowsCI/CD for data and machine-learning codeModel monitoring and auditable controlsMentoring in clean code, testing and sustainable delivery

In practice

Secure pipelines for health data

A small, sharp process

Three steps.
Zero theatre.

Every engagement is shaped around fast learning, visible progress and a clean handover.

01

Find the lever

Clarify the decision, bottleneck or repeated task where data can create the most useful change.

02

Prove the path

Build the smallest credible solution, test it against reality and measure whether it deserves to scale.

03

Automate & transfer

Make the winning path reliable, observable and understandable—then put it in your team’s hands.

Working together

How engagements work

Engagements grow with evidence, not with assumptions. You commit to the next step only once the previous one has shown it is worth it.

01

Start small

Most engagements begin with one focused question: where is the lever? The result is a clear view of where data can create useful change—and where it cannot.

02

Proof of value

If the lever is real, the smallest credible solution is built and tested against your actual data and processes, so the decision to scale rests on measured results.

03

Implementation & handover

The proven path becomes a reliable, observable and documented system. Your team is enabled to operate and extend it, so the result does not depend on me.

04

Remote, in two languages

I work exclusively remotely with clients across Germany, the DACH region and Europe. Working languages are German and English.

Contracts & compliance

The paperwork, settled up front

Clear contractual terms before any data is touched, so your legal and procurement teams can sign off quickly.

  • Confidentiality & data protectionI sign an NDA before sensitive details are shared, and a DPA under GDPR Art. 28 whenever I process your personal data.
  • Your environmentI prefer to work inside your own systems and cloud accounts, so your data stays there.
  • EU AI ActAs part of model governance, I help assess and document AI systems for EU AI Act readiness.
  • InvoicingInvoices in EUR from Dr. Franke Datenwissenschaften OÜ (Estonia), with VAT reverse charge for EU business clients.
  • LanguagesContracts and documentation are available in German and English.
  • InsuranceProfessional liability insurance is in place.

Questions & answers

Short answers to common questions

  1. What does a data science & AI consultant do?

    A data science & AI consultant turns data into better decisions and into systems that run in production. In my work, that covers data and AI strategy, applied machine learning and LLM systems, data platforms and MLOps—from the first assessment of your data to a documented system your team can operate.

  2. How does a project start?

    With a free initial call of about 30 minutes, booked via Calendly or by email. We clarify the decision or process you want to improve and which data and systems already exist. If there is a real lever, the next step is one focused question, not a large commitment up front.

  3. Do you work remotely?

    Yes, exclusively. I work with clients across Germany, the DACH region and Europe, all of it remotely. Working languages are German and English. Wherever possible, I work directly inside your systems and cloud accounts, so your data stays with you.

  4. Who runs the system afterwards?

    Your team does. Every engagement ends with a reliable, observable and documented system and a clean handover, so the result does not depend on me. Where it helps, I mentor your engineers in clean code, testing and CI/CD so they can operate and extend the system with confidence.

  5. How is confidential data handled?

    Under contract and, wherever possible, inside your own environment. I sign NDAs and, when I process personal data, a data processing agreement under GDPR Art. 28. I prefer to work in your systems and cloud accounts so data does not leave them. Past work includes secure pipelines for sensitive health data.

  6. Do you work with mid-sized companies?

    Yes. Engagements start with one focused question and grow only with measured results, which suits mid-sized companies that need value quickly without first building a large data team. The experience comes from production systems for Audi, Porsche and Volkswagen, scaled to the smallest system that solves your problem.

Not sure where to begin?

Bring the messy problem.

Book an initial call