Case studies

Engagements, not endorsements.

We name the clients we are contractually cleared to name and keep the rest anonymised until they are. Here are six current European public-sector engagements and three commercial precedents that established the practice.

Current engagements

Public-sector programmes in flight.

🇩🇰Denmark·HealthActive engagement

Real-world data for precision cancer medicine

Major Nordic national medicines authority

Problem

Ongoing validation of new cancer medicines in real-world patient populations, against fragmented national health registries.

Approach

Privacy-preserving pipeline over LPR3-class inpatient data, pharmacy dispensing, and civil-registry linkage, with reusable protocol templates that compress follow-up cycles from nine months to three.

Capability

Regulatory-grade epidemiology, Git-versioned analytical code, sworn translation, monthly steering cadence.

PythonDuckDBPostGISdbtQuartoGDPR-native pipelines

🇬🇧United Kingdom·Public management · MobilityActive engagement

County-wide traffic and origin-destination analytics

UK county council

Problem

Fourteen local authorities, fragmented ANPR/Bluetooth feeds, and an annual commercial GPS licence the council wanted to replace.

Approach

Doubly-constrained gravity models fused with UK Census and DfT open data, served through an interactive dashboard and REST API. Methodology published on GitHub as open-od-toolkit.

Capability

Open-source replacement for commercial mobility data; no vendor lock-in.

PythonGeoPandasFastAPIH3open-od-toolkit

🇫🇮Finland·Public management · MobilityActive engagement

Real-time traffic situation picture with anomaly alerting

Major Nordic capital city

Problem

City operators needed a unified situation picture for floating-car data with speed-anomaly detection and historical trend context.

Approach

Streaming ingestion of anonymised probe data, statistical change-point detection on corridor speeds, and LLM-assisted incident summaries delivered into the operator console.

Capability

From raw signals to operational insight — within one fused control surface.

LangGraphDuckDBPlotlyWebSockets

🇫🇮Finland·HealthActive engagement

AI and data-engineering panel for a national hospital system

Largest Nordic hospital district

Problem

Multi-site demand for AI application development and lakehouse operations across five regional health authorities, with EU AI Act and clinical-data constraints.

Approach

Joined the supplier panel across two specialisms — AI application development (LLMs, agentic workflows) and data production (Data Vault 2.0 on Databricks) — with healthcare-regulated delivery practice.

Capability

EU AI Act readiness, clinical-data discipline, dual-track delivery.

DatabricksData Vault 2.0LangGraphDelta Lake

🇳🇱Netherlands·Public management · JusticeActive engagement

Criminal-justice data platform for a three-agency consortium

National probation service, partnered delivery

Problem

Consolidate probation-process data for three justice organisations into one analytics platform supporting 200+ dashboards and 200+ custom analyses per year.

Approach

Partnered delivery with an NL-based prime covering Dutch language and justice-sector clearance; we own the data-engineering, AI-insight, and dashboarding layers.

Capability

Delivery in the most sensitive public-sector domains, under clear data-protection control.

PythonPostgreSQLdbtPower BIAirflow

🇩🇰Denmark·Public managementActive engagement

Municipal data-quality framework

Major Danish municipality

Problem

A continuous framework for ad-hoc data engineering, profiling, and quality-assurance work across municipal data estates.

Approach

Registered once on the dynamic purchasing framework; individual call-offs delivered via our open-source data-quality-toolkit with declarative rules and AI-assisted anomaly detection.

Capability

Inspectable, auditable alternative to closed enterprise DQM suites.

data-quality-toolkitPythonGreat ExpectationsDuckDB

Named precedents

Commercial roots.

🇪🇺Pan-European·Financial servicesDelivered

EU RMBS rating models on 2B+ records

DBRS Morningstar

Problem

Rating models for European residential mortgage-backed securities, integrating over two billion records of unstructured loan-level data.

Approach

Document ingestion and extraction pipelines feeding XGBoost-based rating models, deployed via MLflow with full regulatory auditability.

Capability

Document processing at scale, regulatory audit trail, production ML ops.

XGBoostMLflowSparkPython

🇪🇺Europe·Cross-sectorDelivered

Enterprise GenAI and RAG for Industry Solutions Delivery

Microsoft Corporation

Problem

European enterprise clients across energy, biotech, and financial services needed production-grade retrieval-augmented generation over unstructured PDFs and platform data.

Approach

Document ingestion pipelines, intelligent extraction, and enterprise platform integration under Microsoft's ISD delivery umbrella.

Capability

RAG in regulated enterprise settings, multi-sector delivery.

Azure OpenAILangChainAzure AI SearchPython

🇬🇧United Kingdom·Financial servicesDelivered

Revenue forecasting with alternative datasets

ARM Holdings plc

Problem

Strategic foresight on royalty and non-royalty revenue trends requiring integration of external and alternative datasets.

Approach

State-of-the-art time series models fused with alternative-data features; production data pipelines and enterprise-scale backtesting.

Capability

Production data pipelines, alternative-data integration, enterprise scale.

ProphetPyTorchSnowflakeAirflow

Engage

Brief us on your programme.

We work with public-sector buyers in the UK, the Nordics, the Benelux and the Baltics. Tell us the problem — we'll tell you if we can help.

Start the conversation→See case studies→