The Engineering Story

While preparing for conventional academic paths during high school, I realized that the feedback loop of classroom learning was too slow for the scale of my curiosity. I did not want to study the history of computation from a textbook; I wanted to build systems that real users could interact with. I made the choice to step away from the degree path.

To me, a degree was a proxy metric — a sheet of paper certifying compliance rather than raw engineering capacity. I wagered that clean repositories, working deployments, and a relentless builder's mindset would speak louder than any credential. I spent my high school nights in terminal shells, reading documentation, and shipping code, entering the industry directly through proof of work.

I came in thinking frontend was about CSS, components, and good taste. My first production incident corrected that quickly. A modal trapped focus the wrong way and a screen-reader user couldn't escape it. The bug had been live for three weeks.

That single ticket reshaped my mental model. I started reading WCAG instead of skimming it. I started writing semantic HTML before I touched a single class. I learned that the parts of frontend nobody sees are usually the ones that matter most.

I joined a team where four products shared one design language but no shared code. Every product had its own Button, its own Modal, its own date picker — all subtly different, all subtly broken in different ways.

I spent a quarter building a real design system. Tokens, primitives, documented patterns, and a Storybook that engineers actually opened. By the end of the year, the four products felt like one product. That was the first time I understood what leverage means in frontend work.

We had a content site bleeding traffic from slow loads. I treated it like a product problem instead of a technical one. I instrumented Real User Monitoring, mapped the worst routes, and worked backwards from the metric the business cared about: time to first useful paint.

Six weeks of route-level code splitting, font subsetting, image budgets, and a much more disciplined component layer later, the score landed at 96 and bounce rate dropped almost a third. The lesson stuck: performance is a design decision, not a cleanup task.

I was handed a clean directory and tasked with building an ERP-style construction dashboard and its backend. I chose NestJS with PostgreSQL for the server, React Native for the site inspector app, and Docker for containerized local dev.

That year taught me that application logic is only half the battle. Designing secure networks (VPCs) on AWS, writing Prisma migrations that don't lock tables, and configuring container health checks are what keep the system alive at 3 AM.

We started running LLM features locally to bypass cloud API latency. I set up Ollama local instances, created Hugging Face pipelines for text classifications, and built custom Model Context Protocol (MCP) servers to bridge our database context to LLM developer agents.

I also spent time mentoring junior engineers on type-safety boundaries between NestJS and Next.js, and how to write automated GitHub Actions for deploying OpenNext edge configurations. Helping others write resilient code became my main leverage point.

I am ready for the next level: a product team where fullstack and devops are treated as a unified discipline, where data safety and container performance are first-class citizens, and where AI automation is integrated responsibly and securely.

I want to keep building systems — from database schemas to edge configurations and React Native views — that feel snappy, resilient, and inevitably simple.