The tools I actually build with.

Asked about often enough to write down. The theme: a boring, dependable stack that lets AI do the accelerating.

Workstation

  • Dell XPS 15 9530

    13th Gen Intel Core i9-13900H, 64 GB of RAM, and 2 TB of storage. Enough headroom to run local dev servers, browsers, and a couple of AI agents side by side without thinking about it.

  • Ubuntu 26.04 LTS

    Linux is where the tools I deploy to actually live. Developing on the same family of systems I ship to removes a whole class of surprises.

Development tools

  • VS Code

    My daily editor — one consistent home for TypeScript, Tailwind, and the terminal, with the extensions kept to a disciplined minimum.

  • Chrome DevTools

    Where the real debugging happens: network traces, performance profiles, and accessibility audits before anything ships.

  • Git & GitHub

    Every project lives in a repo with issues, pull requests, and reviewable history — including the specs my AI agents work from.

AI engineering

  • Kiro

    Home base for spec-driven development: requirements, design, and tasks agreed before any code is written — in the IDE, the CLI, and on the web.

  • Claude Code

    My agentic pair for implementation, code review, and end-to-end verification. Fast hands, held to engineering standards.

  • MCP servers

    The Model Context Protocol connects the agents to real tools — driving a browser for E2E checks, researching the web, and more — so AI work ends verified, not just generated.

Stack

  • Next.js + TypeScript + Tailwind CSS

    Component composition, type safety, and a consistent, responsive design system — the default front end for everything I build.

  • AWS Amplify Gen 2

    Auth, data, and hosting as code on serverless AWS — fast iterations, low operational overhead, and security by default.