Full-Stack AI Applications Architect  |   Data Scientist  |   Multi-Physics R&D Engineer  |   Ph.D.  |   30+ years

Selected Work

PostalDataPI

Global postal-code validation and enrichment API covering 246 countries + territories (1M+ postal codes) at sub-5ms response. Built agent-native from day one — first-party Python, Node, and MCP SDKs, OpenAPI 3.0, and llms.txt for direct AI-agent consumption. Self-hosted Node 22 / Next.js 15 API surface with in-memory cache; Vercel for marketing, dashboard, OAuth, and Stripe webhooks; Neon PostgreSQL with Drizzle ORM. Transparent single-tier pricing at $0.000028/query. Founder & Owner. postaldatapi.com

Fungaia Coffee Company

Functional-mushroom coffee and cacao D2C brand based in Boise, Idaho. As Director of Technology and part owner, I own the full Next.js e-commerce stack on Vercel — Square (payments, orders, inventory), Shippo (real-time shipping), Resend (transactional email), and Printful (print-on-demand merch) — and advise on unit economics and channel strategy beyond engineering. fungaiacoffeeco.com

Senior AI Consultant at Unify Consulting — Microsoft ABM AI team (current)

Partnering with the Microsoft Account-Based Marketing AI team on the analytics layer for Microsoft's account-based seller recommendation engine. Building reusable analytical infrastructure: Spark/PySpark notebooks against the recommendation-outcomes Delta lakehouse, a curated knowledge base of data snapshots and business references that compounds across sprints, and a self-contained Microsoft Fabric reporting tool that classifies free-text seller dispositions against a canonical taxonomy via Fabric AI Functions (gpt-4.1-mini) over millions of rows. Developed and documented a local-first / Fabric-validated working pattern to keep iteration speed high without sacrificing production rigor.

Integrated Machine Learning & AI (2018–2022) — Founder

Founded and grew a worldwide mentoring community to 65K+ followers through technical and mentoring blog posts, bi-weekly tech chats where I helped data scientists work through detailed technical questions, frequent keynote talks, and ML book reviews for Packt. Starting data science communities inside the companies I've worked for is a pattern, not an accident.

Echo Global Logistics — Outlook AI Add-In for single-shipment quotes (Senior Data Scientist)

Sales reps were receiving 30,000+ single-shipment quote requests per day via email — not managed accounts, so each request consumed sales-rep time for little profit. The software dev team was not available for front-end work, so I developed the full stack: an Outlook Add-In LLM-based AI Web App. Quote-request handling dropped from 5+ minutes to 15 seconds. Meaningful margin recovery for the company and compensation lift for the reps in the field.

ON Semiconductor — autonomous wafer-level vision-test automation (2016)

ON Semiconductor's wafer-level vision test for digital-camera imaging die took 2+ person-weeks per tuning to identify subtle defects other tests missed. After tenured teammates said it couldn't be automated, I built it on my own time in the evenings. Three months later: x/y pixel positions as features, grayscale values as labels, 5th-order feature engineering for a fit that wasn't overfit, statistical-band clustering for pass/fail. The application graded better than I did manually.