# Franklin Javier — Principal Engineer

> I'm a Principal Engineer and I've been doing front-end for 19 years, building fast web and mobile apps. These days almost everything I do involves AI: I build agent harnesses, handle orchestration and write the tooling that makes coding agents reliable.

I work on React and Node.js architecture, performance and observability, and on agentic development: skills, evals and orchestration across Claude Code, Cursor, Codex and Gemini CLI. I also care about accessibility and design systems, so the interface works for everyone.

Outside work, I mentor, pair program and contribute to open source, including AI and agent tooling projects.

## Work Experience

- Principal Engineer — Stone Giant Studio (2023 - Present)
- Principal Engineer — Grupo Boticário (2021 - 2023)
- Head of Technology — Beleza na Web (2015 - 2020)

## Dev Stack

React / Typescript, React Router / Remix / Node.js, Tailwind CSS, Performance Optimization, AI Agents / Orchestration, Design Systems, Accessibility (a11y)

## Contact

- Lisbon, Portugal
- Email: franklinjalves@gmail.com
- <https://www.linkedin.com/in/franklin-javier-98504321>
- <https://github.com/franklinjavier>
- <https://x.com/franklinjavier>
- <https://dev.to/franklinjavier>
- CV: https://franklinjavier.com/cv/

## Recent Posts

- [How an agent loop works](https://franklinjavier.com/blog/how-always-on-agents-work/) (2026-10-11): How always-on agents handle queues, recovery, and context, based on Lee Robinson's Stanford CS146S lecture, followed by my own plan for a first experiment.
- [My old MacBook is now a context server](https://franklinjavier.com/blog/old-macbook-context-server/) (2026-10-08): Shared context across coding agents, machines, and projects: my 2017 MacBook stores session history and summaries, with repository checks before resuming work.
- [Lisbon AI 2026: agents, evals and security](https://franklinjavier.com/blog/lisbon-ai-2026/) (2026-09-25): My notes from all 32 talks at Lisbon AI 2026, with the demos, the numbers and the audience questions.
- [I replaced Tailwind with StyleX. The site didn't get faster.](https://franklinjavier.com/blog/stylex-personal-site/) (2026-08-25): I migrated a site that already scored 100 on PageSpeed to StyleX, measured the result, and found the real performance win somewhere else.
- [We published 28 of our agent skills as open source](https://franklinjavier.com/blog/open-source-agent-skills/) (2026-06-26): Stone Giant Studio's internal skills for AI coding agents are now public - markdown files, no dependencies, portable across Claude Code, Cursor, Codex and Gemini CLI.
- [What web performance actually means in e-commerce](https://franklinjavier.com/blog/web-performance-ecommerce/) (2024-03-20): We took a storefront from 15s to 3s on throttled 3G and conversion went up 45%. What to measure, where the time really goes, and why the biggest wins are usually deletions.

## Side projects

- [Zelus](https://zelus.sh/?utm_source=FranklinJavier&utm_medium=site): Condominium management, with issue tracking and an AI assistant grounded in the condominium's data.
- [Project Campfire](https://projectcampfire.io/?utm_source=FranklinJavier&utm_medium=site): Pulls tasks, decisions and deadlines out of conversations, with owners, due dates and a trail back to the source.
- [Robot Loupe](https://robotloupe.com/?utm_source=FranklinJavier&utm_medium=site): Human-in-the-loop review for AI-generated database changes, checking impact and standards before they ship.

[Blog](https://franklinjavier.com/blog/) · [About me](https://franklinjavier.com/about/) · [Contact](https://franklinjavier.com/contact/) · [Privacy](https://franklinjavier.com/privacy/)

---

Every page on this site is also available as markdown: request any URL with the `Accept: text/markdown` header. See [llms.txt](https://franklinjavier.com/llms.txt) and the [sitemap](https://franklinjavier.com/sitemap.xml).