Difficult things, actually done.

The situation

One of the world's largest aerospace companies needed to see its commercial airline engine fleet as it flew, not in retrospective reports.

The work

Helped build and deliver a cloud data platform analysing petabytes of engine data for patterns and anomalies, on Microsoft Azure and Power BI.

The outcome

A real-time view of the fleet, with the ability to predict engine maintenance events early enough to minimise aircraft time out of service.

Petabyte-scale · Azure · Power BI · Predictive maintenance

The situation

When the pandemic hit, a leading UK university needed roughly 7,000 staff and students working and learning remotely, at speed, on infrastructure never designed for it.

The work

Delivered a modern, scalable desktop platform on Azure Virtual Desktop and enabled the scaled rollout under genuine time pressure.

The outcome

Remote teaching and working at full scale, on a platform designed to outlast the crisis that forced it.

~7,000 users · Azure Virtual Desktop · Pandemic response, 2020

The situation

AAA games studios each held their own analytics, with no central view across titles and platforms, at a time when Azure itself was young.

The work

Centralised studio analytics on the Azure platform, and later dogfooded the internal Microsoft tooling that became public services including Power BI, while migrating Global BI workloads from on-premises to Azure.

The outcome

A single analytics estate across studios, built early enough that the tools it proved out went on to become products.

AAA studios · Early Azure · Power BI lineage

The situation

Significant, unexplained losses in stock management, with the evidence buried in high-volume transactional data.

The work

Designed and built a data warehouse platform to record the patterns and trends in stock movement, with automated analysis around the events where losses clustered.

The outcome

The stock and logistics lifecycle segmented and visible, so losses could be traced to causes rather than absorbed as a cost of trading.

High-volume transactional analysis · SQL Server

The situation

Donated goods were leaking out of the collection network before reaching stores, and the retail side of the charity ran with no regular, accurate analysis.

The work

Designed and built a greenfield Azure data platform giving a 360-degree view of operations, connected legacy POS terminals for real-time sales, and introduced IoT tracking across vehicles, textile banks and clothing bags.

The outcome

Circa £3.5 million saved per year from that work alone.

£3.5m annual saving · Azure · IoT · Real-time retail

The situation

Fixed-odds terminals and spread betting carried fraud patterns the company could only see after the money had gone.

The work

As key and founding member of a consulting partner's Data & Advanced Analytics practice, designed and built a cloud data platform revealing, in real time, who was playing the odds on a particular event, and recruited the architects, engineers and data scientists around it.

The outcome

Fraudulent activity visible as it happened rather than in the post-mortem, and a data practice that outlived the engagement.

Real-time · High-volume transactions · Practice build-out

The situation

An independent economic research business held a deep proprietary archive: over 78,500 publications, roughly 89,000 chart assets and tens of thousands of structured forecast data points. It wanted generative AI to materially change how customers and internal teams worked with that archive, and it needed propositions that would survive contact with production, not demonstrations.

The work

TechGuidr conceived, designed, architected and built both products from scratch: the product thinking, architecture, engineering, data foundations, AI implementation, user experience and delivery. The Research Intelligence Platform lets an institutional user ask a complex economic question in natural language and receive the organisation's current house view, synthesised across the archive with grounded citations and the underlying research surfaced for verification. The Commercial Intelligence Workspace embeds AI into the daily commercial operating process of account teams: portfolio intelligence, renewal risk, structured diagnostics and workflow engines producing action plans with owners and dates.

The outcome

The Research Intelligence Platform went into production; the Commercial Intelligence Workspace proved its value as a working prototype. The ingestion work also surfaced something nobody had asked for: premium research meant to sit behind authenticated access was publicly reachable because of a content-delivery configuration problem. TechGuidr identified it, assessed the business risk and reported remediation to leadership.

78,500+ publications · Hybrid retrieval · Grounded citations · Production

The situation

Workforce Intelligence, a UK technology-enabled learning and wellbeing business, had ambitious enterprise plans: employee diagnostics, organisational analytics and AI-assisted briefings for leadership teams, across an estate that had grown over multiple clouds, services and technology generations.

The work

A deep architectural review found material gaps between the apparent product capability and the underlying systems, including dashboards disconnected from live data and scoring pipelines that did not implement the authoritative calculation methodology. Dan then rebuilt both principal applications from the ground up, re-architected the data platform on an Azure Databricks Medallion lakehouse with verified calculation pipelines and a structured dimensional model, built a context-aware AI Coach grounded in the aggregated analytics, strengthened identity and deployment controls, and carried out a full Azure tenant-to-tenant migration across the business.

The outcome

The platform moved from fragmented, partially disconnected technology to a coherent enterprise architecture, with both rebuilt applications in production. Dashboard queries that had taken four to five seconds returned in under 600 milliseconds once the unnecessary middleware was removed.

Two applications rebuilt · Databricks lakehouse · Tenant-to-tenant migration