Welcome!

Opening

Art is never finished, only abandoned.

This is a phrase that resonates with many software engineers, especially when it comes to personal projects. They are often the first things to be dropped for a higher priority or more interesting proposition - and they’re rarely picked up again.

I’ve had the same relationship with my personal domain, portfolio, digital garden, blog… It has changed form many times and, almost every single time, it was never completed to a point where I was truly happy with it. I’ve always been a little overambitious - which I think is a good attribute - but it does mean my drive fades when I can’t progress as quickly as I’d like with the limited time I have in the evenings and at weekends.

Enter, AI

There’s little doubt that AI will become part of our daily lives in multiple capacities. Its capabilities feel limitless, and its impact will likely be immeasurable as we grow alongside it and discover more good (and bad) applications.

AI has already had a significant impact on the software development industry. Its ability to detect patterns, train on vast codebases, and absorb architectural design principles has made it a powerful partner for engineers. In the last 2-3 years, code-centric models have taken the market by storm - and they’re better than ever.

In my most recent experiments with AI, I’ve been able to:

• Reverse engineer a proprietary UDP protocol from captured packet data and a decompiled, obfuscated Android app.
• Create a personal CRM to help me stay organised.
• Build a travel site using Payload and React for my wife and me.
• Design and implement automation blueprints for GitHub, Home Assistant, and more.

And that’s just my code-related projects - all of which I’ll be writing about soon.

Brief

When I revisited this domain last week, I realised my previous attempt hadn’t gone very far at all. I’d tried to establish a Markdown-based site with the beginnings of a theme - but it didn’t extend much beyond some basic colours and a slightly budget-looking logo.

It was time to roll up my sleeves and finally sort this space out. I had clear intentions:

• Share short-form content (“Bytes”) - quick thoughts and opinions in an X-style feed.
• Share longer-form content (“Megabytes”) - structured stories, how‑to guides, or essays rich in typography and imagery.
• Share photos, ideally daily, along with a story and optional location.
• Build project documentation - sharing problem statements, approaches, and updates for longer-term work.
• Represent my personal brand in a fun, engaging way, while still maintaining a professional space for career prospects.

The brief has lived in my head for years, but I’ve never had the time or momentum to deliver it without bogging down, stalling, and eventually restarting two years later.

Build

Alongside the brief, the technical approach was clear:

• React, using the modern Vite build toolchain.
• Vike for routing and the ability to switch between SSG, SSR, and ISR as needed.
• Tailwind for styling and layout.
• Payload CMS for content creation.
• A headless architecture, separating data and presentation layers.
• Serverless by design, using Cloudflare Workers with D1, R2, and Image bindings.

Using Claude Sonnet 4.5 inside VS Code, I was able to rapidly build out a component library - effectively a living mood board for the site. It includes a colour palette, typography, visual flourishes, and reusable UI elements. It’s a great way to validate whether the theme in your head actually works in practice.

Next, I established schemas in Payload for Byte, Megabyte, and Photo content types. That allowed me to pull structured data via GraphQL and render it across the page templates I had in mind.

From there, the site came together quickly.

I stayed consistent with the visual direction - drawing on core memories from early life. I’ve always found 80s and 90s technology playful, quirky, and full of character. I wanted to replicate as much of that feeling as technically possible, and I’ll continue expanding the component library as the site evolves.

What’s next?

Next up are the Project pages. I don’t want these to follow a rigid template as each project should have its own bespoke style and approach.

For now, clicking on any of the project videotapes will show a “Coming Soon” page. But now that I’ve reached a launch milestone, I expect to tackle these very soon.

After that, I’d like to dive deeper into visual systems and typography - and explore integrations that pull in fitness, listening, and general life data.

Footnote on AI

This site wouldn’t exist without my rampant use of AI - but it comes with caveats.

Without an engineering mindset guiding the prompts, it can generate absolute slop. It becomes lazy, suffers from “context rot,” and will occasionally lose its mind when it hallucinates gaps in documentation for whatever library you’re using.

I experienced waves of frustration with Claude (and other models) throughout this build. Every line of code needed review to ensure it didn’t drift from my architectural intentions or take inefficient shortcuts.

In software engineering, AI still needs engineers - and engineers benefit from AI. It’s an accelerant, not a replacement. It won’t be taking engineering jobs any time soon, but it will reward those who know how to wield it well.