Software, Code & AI

Foredeck.co

A fully custom HTML site I built from scratch as the public face of my wife’s consulting business — designed, coded, deployed, and hosted on AWS.

Foredeck.co began as a simple idea: give my wife’s consulting business a credible, polished web presence as she grew her client base. Instead of using templates or builders, I hand-coded the site from scratch — a return to the kind of HTML work I first learned in the mid-1990s when I was building my own college website.

I designed the layout, built it in a basic HTML editor, and hosted it on AWS, handling DNS, SSL certification, and deployment. The result is a lightweight, responsive site that does exactly what the business needs. For me, it was also a chance to stretch my hands-on development skills and build something meaningful for someone I care about.

Asset Selection Engine (LoanStreet)

The automation system I built at LoanStreet that transformed a slow, manual process into a scalable engine powering the business for years.

When I joined LoanStreet, we were a small team supporting increasingly complex customer requests. The company’s manual asset-selection process simply couldn’t scale with the demand we were seeing. I designed and wrote version 1 of the automation engine that transformed a multi-day analysis into something that ran in minutes.

The early model was built quickly, but it proved highly effective — so effective that as offerings grew more complex, I iterated on it multiple times, improving performance, data sourcing, and business logic. From 2016 to 2020, the engine I built became the core of LoanStreet’s operational infrastructure, forming the backbone of the company’s ability to deliver accurate, rapid offerings at scale.

Weather App

A lightweight weather dashboard I built with my kids using HTML, CSS, JavaScript, free APIs and help from ChatGPT.

This project started as an experiment with my kids: could we build a simple weather app using AI to help with coding?

ChatGPT was extremely helpful early on, giving me the basic HTML, CSS, and JavaScript, walking me through server setup, and helping configure SSL certificates. But once I began iterating, I encountered one of the well-known limitations of AI coding tools: each revision tended to rewrite large chunks of code, rename variables, and introduce incompatibilities that broke previously working sections.

This mean digging in to read through the code, understand what each function was doing, and identify where ChatGPT’s suggestions diverged from earlier versions. After about two hours of back-and-forth, debugging, and small adjustments, I had a fully functioning app that pulled weather and air-quality data from free APIs and displayed it in a rotating dashboard. I kept iterating on it sporadically to improve layout, visibility, and responsiveness.

What made the project meaningful wasn’t the app itself, but the process. It was a great experiment in AI can accelerate development, where it can introduce risk, and why understanding the underlying code still matters. Overall, it was a fun way to take an idea and bring it to life quickly.