The roles changed: engineer, founder, CTO, product leader, engineering leader, principal engineer. The through-line did not. Find the point where software changes the odds, then build across whatever boundaries the product requires.
Infrastructure, ML, consumer products, company building, and the organizations needed to make the work stick.
2019 - Present
Principal Engineer, Agentic Systems
Built at Apple scale, from petabyte data infrastructure to speech and now agentic systems.
2017 - 2019
VP, Product Engineering
Turned a consulting workflow into a graph intelligence product built on billions of public records.
2013 - 2017
First engineer to VP, Engineering
Built an ML sizing platform, stayed through two acquisitions, then led product and engineering globally.
2008 - 2015
Founder, CTO, technical architect
Built products in e-commerce, wine, fitness, and trucking while the underlying markets were changing.
2001 - 2008
Investment Banking Technology
Learned scale by building browser frameworks and compliance systems that other engineers depended on.
These are solo projects. I set the direction, make the product decisions and own the result. A persistent team of agents helps me research, build, test and operate them.
An agent OS for how humans and AI work together.
Voxelbox is the operating system I built for working with agents across everything else I make. Named agents, called stewards, own workspaces, message one another and coordinate work through dispatch, isolated worktrees and scheduled jobs. One Go binary provides the CLI and daemon, so Codex, Claude Code, OpenCode and other agents work against the same tools, memory and trust boundaries. The idea is compound interest: every capability should make the others more useful, and every completed piece of work should leave the system better prepared for the next one.
Solo founder, architect and builder
Built and operated by me, with a team of agents.
Go / Swift / Agent runtimes / Local models
Visit voxelbox.com ↗
An AI-native incubator for AI-native startups.
The incubator I wished existed: one that uses the same agentic systems I build professionally to accelerate the hardest part of starting a company. Founders build their FlightPlan, a complete, evidence-based company foundation, through AI-guided feedback loops. Market analysis, customer discovery, brand identity, pricing, and business model all happen before a line of code is written. I use Maverick to incubate my own portfolio, and every tool I build for it compounds across the others.
Founder, product and systems
.NET / Agent workflows / Product strategy
Visit flywithmaverick.com ↗
A connected hospitality ecosystem built from a 15-year obsession with taste and discovery.
Sotto grew out of Vinelog, my 15-year labor of love. It is now three connected products on one platform: Sommwise for wine education, Ensalle for restaurant intelligence and food-and-beverage operations, and Vinelog for wineries and the people who love their wines. They share one entity system, so enriching a winery for a restaurant list also improves study content and consumer discovery. Vinelog runs in production. I still make every product and operating decision; agents have turned fifteen years of nights-and-weekends work into something I can move every week.
Solo founder, product and engineering
Built and operated by me, with a team of agents.
.NET / Search / Recommendation systems
Visit vinelog.com ↗The operating system for one very serious dynasty fantasy football league.
I built DynastyDB for the CLONE league and have kept it running long enough for the software to become part of the league itself. It handles contracts, salary caps, auctions, restricted free agents, rosters, trades, forums and the administrative work around all of them. I am now moving the live product from its original ASP.NET MVC application to a modern Blazor system without breaking the history or league rules encoded in the old one. It is a useful test of agentic development precisely because the constraints are real: years of data, active users and no room to hand-wave the edge cases.
Solo founder, product and engineering
Built and operated by me, with a team of agents.
.NET / Blazor / SQL Server / SignalR
Visit dynastydb.com ↗Smaller probes where the boundary is still moving. The work is real; the shape is not settled yet.
2025 - Present
Generative creativity tools.
Exploring and experimenting with generative speech, music and images. Building tools that make generative media accessible to creators, starting with myself and my agents.
Swift / MLX / Open models
2025 - Present
The intersection of gaming, crypto and machine learning.
Building on the thesis that games can be a viable system for collecting ML training data.
Unreal Engine / Machine learning / Game systems