Building TrueTrack, a financial analytics platform for automotive and powersports dealerships, at a Jackson Dawson subsidiary.

• Modernizing core systems (GL account-mapping, file-intake) within a legacy data platform
• Coordinating multiple parallel AI coding-agent sessions against a shared codebase to accelerate a large-scale migration
• Building explainer tools and interactive demos to make complex data-modeling tradeoffs legible to cross-functional stakeholders

Partner at a Detroit fintech building invoicing and payment processing for SMBs via online banking.

• Architected Autobooks Hub — an in-browser analytics platform consolidating Receivables, Capital, Accounting, and platform events into a single interface
• Designed the full data pipeline: SQL Server → Azure Functions → Arrow IPC → Parquet → DuckDB-WASM, enabling instant client-side queries with zero server round-trips
• Built a Claude-powered AI assistant that generates SQL from natural language; queries run entirely client-side, data never leaves the browser
• Implemented a 122-permission scoped access system (verb:resource@scope) managed through Auth0 JWT tokens
• Designed a Segment Protocols tracking plan with 150+ events across the full Hub user journey

Owned product analytics and growth engineering, building the data infrastructure to measure and optimize the SMB onboarding funnel.

• Drove 3x+ acquired users, 5x+ active users, and 10x+ gross merchandise volume YoY
• Integrated Autobooks with Segment CDP, establishing the event tracking foundation that now powers Hub analytics

Partner — joined at $15M ARR, departed at $100M ARR. Owned the full marketing automation stack as both engineer and power user.

• Built complex multi-channel automation workflows processing millions of emails daily
• Created developer education content including API tutorials with Serverless, AWS Lambda, and IoT hardware

Early employee at a referral marketing platform; built and scaled growth and marketing through acquisition.

• Integrated referral tracking with Salesforce, SendGrid, MailChimp, PayPal, and Dwolla
• Owned the full marketing data stack: attribution, analytics, and automation

SaaS platform analyzing environmental and operational data to predict retail location revenue performance.

• Built data pipelines ingesting from Google Places, Yelp, Foursquare, Facebook, and weather APIs into a unified analytics engine

A declarative framework that splits applications into pure data schemas ("edges") and pure functions between them ("nodes"), wired together in a topology declared up front — a bet on structuring software for a world where agents draft and read as much code as humans do.

• Failure is a typed edge, not an exception — an unhandled case is a type error, not a 3am surprise
• A decision becomes a type: a classifying node's output type is the proof a decision was made, so illegal states become unrepresentable

github.com/jordanskole/weir →

A Michigan vacant-land data layer built to compare parcels on physical and regulatory attributes — architected so the system only ever emits measurements, never verdicts, keeping judgment calls separate from the data itself.

• An MCP server boundary that structurally cannot return a verdict — the separation is architectural, not a convention

github.com/jordanskole/blue-ribbon-properties →

SQL against the entire US banking system, running in your browser tab — FDIC, FFIEC/NIC, and NCUA regulatory data loaded into a client-side DuckDB engine, queryable in plain English via an AI agent or directly in SQL.

• One schema drives DuckDB DDL, TypeScript row types, join graphs, and LLM prompts — the ETL, web app, and AI assistant can never disagree about what a column means
• Every insured bank and credit union, every branch, every merger and failure since 1994, joined across sources rather than siloed per-dataset

github.com/bankql/bankql →