About

AI-expert data teams · player/coach · 25+ years

I'm Dave Holmes-Kinsella — a data and analytics leader and hands-on builder. For 25+ years I've been turning messy data into decisions, products, and AI that holds up, across healthcare, fintech, and telecommunications.

I lead as a player/coach. I build the sources of truth teams can trust, the engineering practices that keep them trustworthy, and the analytics teams that run on them — and I stay close enough to the work to ship the first version myself. That's meant reconciling more than $100B in ledger activity to the penny, cutting a 14-day month-end process to a minute, and putting agentic systems into production.

Right now I'm focused on what it takes for an analytics organisation to work AI-natively — not the hype, but the evaluation loops, standards, and habits that make AI dependable when the tools change faster than the org charts do.

I write about data, AI, and the people caught in the middle of it all. If something here was useful, I'd love to hear from you.

What I'm Good At
AI-native teams Evaluation loops and AI-assisted development standards; agentic systems in production (5.5× call completion)
Sources of truth Ledgers and reporting that reconcile to the penny, with explicit ownership and upstream monitoring
Teams & standards Building and upskilling analytics teams; architecture, testing, and code-review standards
Data as product Turning internal analytics into customer-operated products (+267% platform adoption)
Selected Work
wingman

Career-intelligence platform with an MCP server, multi-provider inference, and source-preserving retrieval.

work-ledger

Analyzes work done with AI and flags repeated processes better built as deterministic software.

tricorder

Captures, measures, and operationalizes the knowledge created during code review.

The Life Expectancy Misconception

Seven interactive views into what 'average life expectancy' actually meant across history.

Tab Tamer

Sort, search, group, and close Chrome tabs from the toolbar. No backend.

Stack
AI Claude Code, MCP, LLM orchestration, prompt & context engineering, NotebookLM
Languages SQL, Python
Data dbt, DuckDB, Pandas
Database Basically all of them
Visualization D3.js, Vega-Lite, React, Tableau, Metabase, Sigma