Aaron Robbins
The Cascadia Portfolio
Cascadia is a portfolio built to be checked, not taken on faith. Each module carries its own governance layer — certified measures, frozen baselines, a validation suite that fails the build rather than ship a number I can’t defend — so a claim on this site is a claim you can verify against the data behind it, not just read. The stack varies by problem: a governed SQL Server → Microsoft Fabric → Power BI enterprise pipeline, pragmatic local SQL Server + Power BI, lightweight Python → DuckDB/SQLite → static ECharts on real SEC, federal court, and economic data, and the program side of BI — platform migration and metric governance. The through-line isn’t the architecture; it’s that every number on every page can be tested against its source.
Featured Case Studies
Revenue Operations Analytics
Cascadia Revenue Assurance
NewLive
What a customer is contracted for and what they are billable for are different numbers. Three asymmetric timing rules — adds land at once, reductions wait for the term boundary, cancellations ride out the term — keep the two apart for up to a year, with nothing erroring and no register showing it: 5,875 licences at June 2026. Derived twice, by two engines written to be different, agreeing on every cell. Synthetic data throughout.
Legal Operations Analytics
Cascadia Fee Examiner
NewLive
A bankruptcy research database, a public court docket, and two task-code vocabularies — three sources, three owners, none sharing an identifier. A stated rule resolves 9 of 25 caption-to-database firm pairs; the other 16 are published with a reason, not dropped. Reported at its actual state: 48 of 216 eligible cases searched so far.
Legal Operations Analytics
Cascadia Matter Ledger
Live
Eleven million public federal civil case records, frozen and hash-verified, modelled into certified legal-operations measures — then a scheduled pipeline that re-asserts its own invariants twice a day, reconciles against the frozen baseline, and publishes the failures. Ask it one ordinary question and the ungoverned answer comes back negative.
Supply Chain / Distribution-Centre Governance
Cascadia Control Tower
Live
Three fill rates computed from one dataset, all arithmetically correct and 7.0 points apart — and a split-shipment cost that none of them measures. Certifies one definition, prices the leak against a modelled no-split counterfactual, and shows why no single split threshold works across two banners. Synthetic operations, anchored to BLS, Census and SEC data.
Pricing Governance / Financial Analytics
Cascadia Deal Desk
Live
Multi-quarter pricing agreements as a governed dataset rather than a signed PDF — every quote line matched to the agreement in force on its own date, the gap priced in dollars, and a materiality threshold you calibrate before you automate an alert. Synthetic data throughout.
Financial Analytics
Cascadia Finance
Live
FormFactor (NASDAQ: FORM) GAAP financials from SEC EDGAR XBRL — an executive close pack in Python → SQLite → static ECharts, with frozen-snapshot governance and explicit GAAP-vs-non-GAAP labeling.
BI Leadership / Governance
Cascadia BI Migration
Playbook
A 24-week program to migrate an organization off legacy BI onto a governed, AI-assisted platform — certified metric layer first, parity QA as the deliverable, wave-by-wave sunset. An illustrative program design.
Healthcare / Staffing Marketplace
Cascadia Staffing
Complete
Contingent-labor economics for 14,362 U.S. nursing facilities on CMS payroll data — 21M staffing records, contract vs. employee hours by role and day, with a certified-measure governance layer validated against independent SQL.
Healthcare / Pharmacy
Cascadia Pharmacy
Complete
GLP-1 spending trends (CMS Medicare Part D) and adult immunization coverage (CDC), with a messy/incomplete-data cleaning showcase.
Manufacturing
Cascadia Medical Devices
Complete
A dimensional model over 5.9M production events plus real NASA turbofan sensor data, tracing a root-cause story from station degradation to on-time delivery.
Retail
Cascadia Clothing
Coming Soon
Demand, merchandising, and fulfillment analytics for a retail apparel scenario.
Beyond the Portfolio
Mathematics · A pet project
The Combinatorics of Sibling Conflict
Preprint
Not an analytics module — a short mathematics paper. The number of ways n children can take sides against each other, with the rest uninvolved, is (3n − 2·2n + 1)/2, the Stirling number S(n+1, 3). Of the 25 ways four children can, 6 are one against one; of the 3,025 ways eight children can, 28 are.