Things I've
actually built.
Most of the production software I've written belongs to former employers and can't be shown publicly. Here's a representative sample of what I can show, and the thinking behind it.
TableGrade.org
Restaurant health inspection trend summaries powered by the Anthropic API, used by Clark County residents.
The problem
Clark County publishes health inspection records publicly, but the raw data is difficult to interpret. A restaurant might have had one bad inspection three years ago and ten clean ones since. A raw score doesn't tell you whether a restaurant is on an improving or declining trajectory.
What I built
TableGrade pulls inspection records, feeds them to Claude via the Anthropic API, and generates plain-language summaries describing each restaurant's health and cleanliness trend over time. The summaries are specific, not generic boilerplate, because the model is working with real data about that specific establishment.
The result
A public tool that people actually use to decide where to eat. The kind of thing that would have taken a team weeks to build five years ago, built by one engineer in a fraction of that time because the AI does the hard interpretive work.
The lesson
This is what small, well-scoped AI applications look like when they work. The AI isn't doing everything, it's doing the specific thing it's actually good at: synthesizing information into readable summaries. The rest is regular software.
Visit TableGrade.org →Payment Processing Infrastructure
Feature delivery and systems work at Webconnex, a bootstrapped payment SaaS handling millions of transactions for events, nonprofits, and ticketing clients.
The context
Webconnex builds payment processing and event registration software used by tens of thousands of organizations. The engineering team is small by design, which means every engineer carries significant surface area and works close to the customer.
What I shipped
Multiple full-stack features from concept to production: registration flow improvements, payment infrastructure updates, reporting tools, and internal tooling. All in a codebase that handles live financial transactions for paying clients, with actual money on the line.
What it taught me
Fintech at a bootstrapped company is where you learn to make hard tradeoffs under hard constraints. Not "what's the best possible solution?" but "what's the best solution we can build, maintain, and support with the team we have, by the deadline that actually matters?"
That mental model, calibrating ambition against capacity and timeline, is the most valuable thing I bring to consulting engagements. I don't recommend things that don't fit the reality of your situation.
Walmart Global Tech
Staff-level engineering work on consumer-facing systems serving tens of millions of users. The scale discipline most companies never develop.
What operating at Walmart scale means
When you build software that millions of people use simultaneously, the consequences of careless decisions are immediate and visible. You develop an instinct for which shortcuts will cost you later and which are fine. You learn that "it worked in testing" is the beginning of the story, not the end.
What I bring from it
Discipline. Rigor. A clear-eyed view of what can go wrong in production and how to design systems that fail gracefully when it does. This is overkill for many small business applications, but it means I know where the corners are, and I make conscious decisions about when it's okay to cut them.
Want to add your project here?
Let's build something worth showing.
Start with a free 30-minute call. I'll tell you what I think is achievable and what it would take to get there.
Book a free consultation