These are demonstration projects built on fully synthetic data to show method, not client information. Each one is designed around a real operating question in a specific industry, built end to end (data generation, modeling, DAX, Python and R, page design), and documented so a technical reviewer can reproduce the numbers. Client work is available on request under NDA.
Power BI · Operations
Voyage & Cargo Control Tower
An operations control tower for an ocean carrier: a balanced scorecard for routes (financial, customer, operations, risk, with what-if weights), plan-versus-actual against port traffic and weather, near-real-time vessel and container status on an Azure Map, stevedoring load and unload progress, reefer telemetry, tariff and landed-cost tracking, and transit-time tiers (average, good, better, best) with factor-lift attribution.
Techniques: star schema with 18 synthetic tables, calculation groups, field parameters, what-if parameters, SVG measures, decomposition tree and key influencers, Python visual, drill-through and cross-report drill-through, Activator alerts, write-back via user data functions.
Power BI · Sales analytics
Sales Intelligence & Forecasting Studio
A sales cockpit for the same carrier: revenue and year-over-year by service and customer, a customer-health and churn-risk score, share-of-wallet estimates, quote win rates and loss reasons, a logistic price-response curve with what-if pricing, a six-month volume forecast with back-tested error bands, and a Customer 360 drill-through page with a write-back “log touch” action.
Techniques: Python script step in Power Query for forecasting (Holt-Winters plus ridge ensemble), Python and R visuals, calculation groups, bookmarks, dynamic RLS and OLS, TMDL, Fabric notebook for the medallion layers.
Power BI · Finance
Lane Profitability & Price-Volume-Mix
Margin by trade lane with a price, volume, mix and cost bridge against budget, container fill-rate economics (what each point of fill is worth), and a calculation group that applies time intelligence and currency formatting across every measure without duplicating DAX.
Fabric · Machine learning
Roll-Risk Early Warning
A Fabric medallion pipeline (bronze, silver, gold) feeding a logistic-regression roll-risk score for LCL shipments, with an exception worklist, a back-tested lift table, EDI 315 feed quality checks, and a user data function for analysts to log the action they took on each exception.
All figures in these projects come from generated data with fixed random seeds; they describe the method and are not real company results. Decks, technical guides and reproducible kits are available on request.
