LinkedIn / selected work
Projects with a decision behind them.
A curated set of case studies from business intelligence, automation, enterprise BI and applied AI. Company details and source data are anonymised where needed.
How to read the numbers
Evidence before adjectives.
Featured cases
4 casesFrom four reports to one operating model.
How a fragmented reporting setup became a Power BI product used for Mexico revenue decisions.
4 → 20+ automated reportsRead case study ↗02 / workflow architectureUnverifiedSix reporting cycles, one repeatable pattern.
An anonymised commercial reporting suite that turns recurring enterprise exports and spreadsheets into reviewable outputs.
6 pipelines · 2 markets · 5+ retailersRead case study ↗03 / operating systemUnverifiedA retail workflow, not just another dashboard.
An anonymised local-first operating system that brings recurring retail decisions into one reviewable workspace.
Reporting · forecast · stock · QA · communicationRead case study ↗04 / shared architectureDocumentedContributing to a BI model that travels across markets.
A scoped contribution to a shared multi-market Power BI/PBIP environment, described without claiming ownership of the whole model.
177 measures · 2 selector tables · 2,181 TMDL linesRead case study ↗Automation & operations
4 casesReporting that runs while the business moves.
A production reporting workflow designed around scale, validation and less manual repetition.
348 stores · 30 countries · 11 checksRead case study ↗06 / retail automationMeasuredFrom daily exports to a reviewed HQ report.
An anonymised retail reporting workflow that turns sales and ticket exports into a consistent, quality-controlled output.
59 stores · 9 products · 45 → 3 minRead case study ↗07 / decision supportMeasuredA weekly commercial pack built around the decision.
An anonymised weekly workflow for forecasting, replenishment and allocation decisions.
~4 h baseline · ~2 min run · ~10 min reviewRead case study ↗08 / stock operationsUnverifiedStock reporting that makes exceptions visible.
An anonymised stock workflow built around reconciliation, reusable mappings and visible QA.
~2–3 h baseline · ~5 min run · 3× weeklyRead case study ↗Forecasting & applied AI
3 casesSeparating the base from the event.
A commercial forecast designed to distinguish underlying sales from the lift created by commercial events.
MTD forecast · base vs event liftRead case study ↗10 / enterprise BIDocumentedA reporting model becomes a product.
Semantic modelling, PBIP/TMDL documentation and Power BI architecture for multi-domain commercial reporting.
114 tables · 235 measures · 25 RLS rolesRead case study ↗11 / AI systemsDocumentedBuilding intelligence under live constraints.
An autonomous trading agent, observability layer and multi-agent development workflow built during the Claude Hackathon Madrid.
Top 10 of 18 · 1,377 testsRead case study ↗The wider practice