Engineering Leader Hands-On Systems Architect
25+ years of SaaS platform engineering across high-volume ecommerce, healthcare, and data-heavy cloud systems. I take sprawling, failure-prone codebases and turn them into simple, message-driven architectures — then coach the engineers who own them next.
Open to senior engineering and engineering-leadership roles — remote (US) or Bellingham, WA.
How I work
I've spent my career alternating between being the architect on the hardest problem and being the manager who makes a team able to solve it. Most recently I did both at once.
Track one
I go after the systems everyone is afraid of — the nightly job that fails every month-end, the service that can't serve your biggest customer. I simplify rather than add, and the result is usually cheaper, faster, and smaller than what it replaced.
Track two
I founded ClearDATA's first Data Engineering team and stayed the principal architect on it. I lead by raising the technical bar — setting direction, coaching engineers through it, and staying close enough to the code to be useful.
Selected work
Every number below is something I can walk you through — the constraint, the decision, and what it cost or saved.
ClearDATA continuously assesses customer cloud environments for HIPAA compliance and remediates them automatically. The platform behind it was a sprawl of single-purpose Lambdas that recomputed in-memory joins and rollups against DynamoDB on every page view. For the largest enterprise accounts, the data volume meant the application simply did not work.
I founded the company's first Data Engineering team, set the architectural direction, and delivered a replacement with a team of senior engineers: an event pipeline that pre-computes results as source data changes, backed by PostgreSQL. It became the foundation of ClearDATA's data and AI capabilities.
Month-end billing failures had forced fire drills for the finance team for years. The logic lived across roughly ten mixed Node.js, Python, and Go repositories. I conceived, designed, and led a ground-up rearchitecture down to three repos, with a single Python monorepo holding about 80% of the system.
Invoice and financial-calculation costs fell from $31K to about $8K per month. The nightly run went from ~10 hours to ~5 — while onboarding more customers and adding Microsoft and Google cloud data to a previously AWS-only system. More data, more processing steps, less compute. The chronic month-end failures stopped, and the architecture was later forked to become the foundation of the flagship product.
I led the design, build, and operation of Dell.com Membership Services as online revenue grew 50× over five years, and architected globally redundant infrastructure that held unplanned downtime to a few hours per year. I owned hardware, OS, and capacity-planning decisions, and internationalized the platform across 14 EMEA and Asia-Pacific countries.
I also led the team that earned Dell's first online customer-profile patent — US 6,862,612, for "express buy" purchasing profiles spanning multiple storefronts.
DebateFrog stages structured, evidence-backed debates between large language models. I built it on a shared TypeScript debate engine with Next.js and React, wired multi-model access through OpenRouter, and added live evidence retrieval via Tavily and Brave search. A model-evaluation suite compares output quality across models, with a public results page; users can swap models in-app for side-by-side comparison.
I operate the whole stack myself — Firestore, Docker containers on Google Cloud Run, CI/CD through GitHub Actions — and develop daily with Claude Code and OpenAI Codex across architecture, implementation, and testing.
Building now
Since April 2025 I've been designing and shipping educational web applications solo, using AI-assisted development end to end.
An educational app that stages structured, evidence-backed debates between large language models — with live evidence retrieval and a public model-evaluation results page.
debatefrog.comNext.js · React · TypeScript · OpenRouter · Tavily · Brave Search · Firestore · Cloud Run
The shared TypeScript debate engine underneath, published as a versioned library module and consumed by two separate applications — DebateFrog and a professional debate workspace.
TypeScript · monorepo packages · versioned library · Docker · GitHub Actions
Toolkit
Contact
If you have a platform that's outgrown its architecture, a team that needs a technical bar raised, or a role that wants both — I'd like to hear about it. I read every message.