Seven Projects.
70,566 Lines of Code.
One Developer.

These projects were built by a single developer, a medical student and military officer with no formal CS degree, using AI as a force multiplier. Claude Code's autonomous agents researched APIs, generated components, debugged edge cases, and iterated on designs in real-time. The result is a collection of production-grade applications that would traditionally require a team of specialists. All were started after March 15th, 2026, built in free time after clinical rotations and in between study sessions. That alone says more about where the tools are heading than anything about me individually.

The motivation behind each project is personal. The NCAAT Bracket, Masters Tournament Hub, and Detroit Sports Hub exist because I love sports. March Madness, the Masters, and Detroit baseball and basketball are things I follow every year with genuine passion, and building interactive tools around them was a natural way to test what AI-augmented development could produce. The Detroit Sports Hub pushed the boundary furthest: 9,500 lines across two broadcast-quality trackers, one for the Tigers season and one for the Pistons' historic playoff run from 14-68 to #1 seed. The Tigers tracker pulls from four public APIs with a news curator that cross-references headlines against actual game results. The Pistons tracker features a live playoff countdown, Cade Cunningham injury watch with recovery progress bar, a 2004 championship tribute comparing that squad to this year's roster, and fact-checked data validated by four autonomous research agents. Together they form a unified Detroit Sports hub with iframe isolation to keep both apps running independently with zero conflicts. The Step 2 CK Study Engine exists because I needed it. No commercial platform offered the adaptive, data-dense study architecture I wanted for my medical education, so I built one. The Patient Handoff Tracker exists because patient handoff is where the majority of preventable medical errors originate, and I wanted to build a system that enforces the evidence-based I-PASS framework at the point of care. Claude Code's /teach-me skill became a built-in tutor, letting me learn new patterns in context as I developed each platform.

Beyond writing code, the AI ecosystem extends into thinking and decision-making. I use /council to pressure-test ideas from multiple angles before committing to an architecture. /brainstorm for structured ideation when I am stuck. Auto-agents to delegate research, refactoring, and testing to parallel workers while I focus on design. These are not gimmicks. They are workflows that change how a solo builder operates.

As a military officer, I also care about understanding how emerging technology reshapes the cyber landscape. I have started incorporating tools like /threat-model, /security-review, and /after-action into that learning process. Knowing how AI systems work, how they can be directed, and where they are vulnerable is no longer optional for anyone responsible for operational security. Building with these tools is one of the best ways to develop that intuition firsthand.

But the deeper thesis is about what comes next. We are approaching a point where AI capabilities compound faster than any individual can fully track, which is exactly why I built the AGI Capability Tracker. It monitors progress across 8 domains, 25+ frontier models, and dozens of benchmarks in real time. The data tells the story plainly: competition math went from unsolvable to perfect scores in 14 months. Coding benchmarks that stumped every model a year ago are now cleared at 94%. Reasoning tests designed to be AGI-proof are already being matched on their second iteration. A restricted model that fewer than 40 organizations can access surpasses human baselines on autonomous computer operation. GitHub Copilot normalized AI in the editor. OpenAI's Codex proved that natural language could generate working code. Model Context Protocol plugins are connecting AI to every tool and API imaginable. These capabilities will not stay behind developer consoles forever. They will leak into every profession, every workflow, every industry. The professionals who thrive in that environment will not be the ones who resist AI or passively consume it. They will be the ones who learned to operate fluently in a dual ecosystem, human and machine working in concert. The bottleneck is no longer technical skill. It is imagination, domain knowledge, and the willingness to build.

These projects are my small proof of that alignment. Evidence that someone who understands the problem space and has learned to work alongside AI can ship real software that solves real problems, in days rather than months. I am still learning, still building, and still a long way from where I want to be. But I believe this is the direction the world is heading, and I would rather be early than late.