I build data and AI solutions.
I fix programmes that aren't landing.

Data platforms

I architect and build data platforms on Microsoft Fabric, Databricks and Snowflake: medallion lakehouses with dimensional warehouse design, data marts and semantic models. I implement the design myself or work with your engineering team, then hand over with technical architecture and operations guides, because running it should be your team's job and not a dependency on me.

Where a platform isn't yet chosen, or needs validating, I run a consultative process covering current capabilities, business data domains, operating model and business goals before recommending one. Several organisations came to me with the platform already chosen and left on a different one. The questions I work through are here.

I build AI products too, so the platforms I design are meant to feed them: governed, versioned data, with PII handled before it reaches a model.

AI data products

The data layer an AI application runs on, built inside your analytics platform and not beside it.

I build AI data products directly into the platform, applying the same lakehouse architecture, data engineering and privacy principles as any other data product to RAG: chunking, embedding and governed retrieval. The model gets data that is governed, versioned and joined to your master data, with PII handled before it reaches a model.

It works on the warehouse you already have, with no rebuild. It starts with a readiness assessment: can your warehouse feed an AI application safely, and what has to change first.

Agentic AI workflows

Workflows that run real operational work, with a human in the loop on the decisions that matter. Not a chatbot, and not a pilot that never leaves the demo.

I have built and run these myself, including outreach and qualification: the system does the research, the drafting and the triage, and I approve what goes out. I run the same pattern over my own software engineering, where a ticket goes through implementation and review as an agent pipeline. I built Relia AI with that pipeline, from zero to App Store approval.

A masters in business analytics, a computer science degree and 25 years of delivery. I can model the process, size the value and ship the system.

Now

I build with AI tooling every day and most of what I know about making it work in production comes from doing it: agentic workflows, RAG, the data architecture underneath, and the judgement AI doesn't have.