Stop waiting three days for a number you already own.
I build the analytics layer that holds an organisation's operational data — then put language models on top of it, so people can ask questions instead of waiting for a report. Independent practice, working with clients across time zones.
Three things I get asked for
Most engagements start with the first and grow into the others.
Analytics engineering
Consolidating operational exports nobody can join into a store you can query, then building the daily, weekly and monthly reporting on top of it.
Claude application delivery
Language models working over your own data — including the caching, model tiering and monthly accounting that decide whether it stays affordable.
Signal & anomaly layers
Detection built on robust statistics rather than naive thresholds, drafting plain-language notes for a human to review before anything is sent.
From BI reporting to a production Claude application
A teleradiology client had hard turnaround commitments and no reliable way to see breaches as they developed. I built the consolidated data layer, three automated reporting cadences, an anomaly detector tuned against a backtest, and a natural-language interface over the whole thing — running at roughly half the inference cost it started at.
Why a BI background matters for AI work
A language model is only as useful as the data you can put in front of it — and most organisations cannot put their own data in front of anything. Years of building reporting teaches you where that data actually lives, why nobody can join it, and which numbers people genuinely decide on. That is the hard part. The model is the easy part.
Have a question your data cannot answer yet?
Tell me what you are trying to see. I will tell you honestly whether it is worth building — including when the answer is no.
Copyright © 2026 | WordPress Theme by MH Themes