What I build

Projects

I choose the medium that fits the problem. Consumer-facing? Lovable or Claude Artifacts. Client workflow? Built inside their existing stack. Missing infrastructure? Custom connector. The goal isn't to build for the sake of building. It's to solve.

01 Live

The Silicon Trail

The startup world needed a mirror. I built one.

The Silicon Trail is a satirical adventure game in the spirit of Oregon Trail, but the wagons are pitch decks and the dysentery is dilution. It exists because the human cost of "grow at all costs and IPO or bust" rarely gets named honestly. Games can say things blog posts can't.

Built as a consumer experience designed for reach, not clients.

Visit the-silicon-trail.com →
02 Live

Get Unboxed

Most people with 20 years of hard-won expertise don't think of themselves as founders. They should.

Get Unboxed helps experienced professionals turn a lifetime of domain knowledge into a business concept and launch plan in 20 minutes. Built for the career pivoter, the corporate veteran ready for something different, and the person doing valuable work for free who hasn't yet been told they could get paid for it. Designed to lower the barrier to building for people who didn't see themselves in the founder story.

Built in Lovable because the audience needed frictionless, not technical.

Visit get-unboxed.lovable.app →
03 Live

Ashby + Claude MCP Connector

Evaluating candidates against a hiring rubric is high-judgment work. The risk isn't doing it wrong. It's doing it inconsistently.

I built a custom MCP connector linking Claude directly to Ashby, authenticated via Google OAuth so only the client's team can access it. Using real hiring manager feedback and resumes, I synthesized a position-specific green flag rubric that runs against every incoming candidate, individually or in batches. Results write directly into sortable Ashby fields and the candidate feed, so the output lives where the team already works. The system also flags when a candidate is misaligned to one role but a strong fit for another open position.

Anyone at the company can download and run it. A batch of 500+ resumes costs under $10 in API calls.

Humans make all the hiring decisions. AI distills the signal so they can make them well.

04 Coming soon

Pre-flight

Enormous attention goes into executing searches. Relatively little goes into making sure the business is ready for one.

Clarity on the specific numbers a hire needs to deliver. Capacity from the founder to actually invest in the process. Realistic expectations of what one person can do. Landmines in the org that will trip them up before they find their footing. These aren't edge cases. They're the conditions that determine whether a search succeeds before sourcing ever starts.

Pre-flight is a diagnostic that identifies likely failure modes before a search launches. Not a gatekeeper. A forcing function for clarity. Founders work through a structured conversation, surface what's unresolved, and get a search brief their recruiter actually has enough to work with.

Building the plane while flying it is fine. Ignoring common pre-search failure modes means you'll spend the search fighting fires instead of making progress.