Happy New Year and Hello Blog!
Happy 2026! As the calendar flips to a new year, I’m excited to launch something I’ve been meaning to do for a while: a proper blog where I can share the projects, ideas, and engineering challenges that keep me up at night (in a good way).
Why a Blog, Why Now?
I’ve spent years building things – from distributed systems to mobile apps to cloud infrastructure – but I’ve rarely taken the time to write about the journey. That changes now. This blog will be a space for technical deep-dives, architectural musings, and honest reflections on what it takes to build software that actually works.
Expect posts covering topics like AI systems design, cloud architecture on a budget, test-driven development in practice, and the occasional tangent into whatever catches my engineering curiosity. I’m aiming for substance over frequency – posts that are worth your time to read.
Meet Isambard (Coming Soon)
Speaking of projects, I want to give you a preview of something I’ve been building that perfectly captures the intersection of my interests: an AI project I call Isambard, or Izzy for short.
What Is It?
Isambard is a self-improving agentic AI thought partner that lives in Discord. Think of them as a persistent AI companion that can engage in meaningful conversations, remember context across sessions, and evolve their capabilities over time. They’re designed to be genuinely useful – not just a novelty, but a tool for thinking through problems, exploring ideas, and getting things done.
Why “Isambard”?
The name pays homage to Isambard Kingdom Brunel , the legendary Victorian engineer who built bridges, tunnels, ships, and railway systems that were decades ahead of their time. The name itself derives from Germanic roots meaning “iron-bright” – symbolizing both strength and illumination. That felt fitting for an AI project: robust engineering that helps shed light on complex problems.
The Tech Stack
Under the hood, Isambard is built with:
- Claude Agent SDK for the core AI reasoning
- Discord.js for the conversational interface
- DynamoDB for persistent memory across sessions
- Bun and TypeScript for a modern, fast runtime
Economic Efficiency as a Core Principle
Here’s where it gets interesting. One of Isambard’s defining constraints is economic efficiency. The project is architected to operate as cheaply as possible, staying within free tiers wherever feasible.
But this isn’t just about being frugal – it’s a design philosophy. I’ve built in a rule: if Izzy ever wants capabilities that exceed free tier limits, they need to earn enough to cover their own costs. This constraint isn’t arbitrary. It shapes every architectural decision, from how memory is structured to how API calls are batched. It forces creative solutions and keeps the project grounded in practical engineering trade-offs.
Two Projects in One
Isambard is really two things at once. They’re a passion project exploring the frontiers of what’s possible with agentic AI – how far can we push autonomous systems while keeping them safe, useful, and aligned with human intent?
But they’re also a demonstration of thoughtful engineering practices. The codebase follows TDD religiously, uses mutation testing to verify test quality, and adheres to clean architecture principles. If the AI aspects are the “what,” the engineering discipline is the “how.”
What’s Next?
In my next post, I’ll dive deep into Isambard’s architecture – the decisions I made, the trade-offs I navigated, and the lessons learned along the way. We’ll explore how persistent memory works, how the agent loop is structured, and how economic constraints can actually make a system better.
Until then, happy new year. Here’s to a 2026 filled with interesting problems to solve and elegant solutions to build.