Full-stack engineer in constant growth — I focus on reliable APIs, validation-first backends, and modern React/Node apps. I like to ship things that are safe to change and easy to run.
I build and maintain backend and frontend systems with an emphasis on clarity, validation, and maintainability. My work centers on TypeScript, Node, React, and related tooling so that apps and APIs stay consistent and predictable.
- Work at Future Secure AI
- Personal projects
- Tech & tools
- How I work with AI agents
- Stats
- Recently watched (YouTube)
- Get in touch
I contributed as a full-stack engineer across the Future Secure AI platform (Sep 2025 – Feb 2026), focusing on reliability, security, and validation so that systems are safer to run and easier to release. All work product and intellectual property from this engagement are the sole property of Future Secure AI.
| Metric | Value |
|---|---|
| Signal commits | 579 |
| Pull requests | 226 |
| Merged PRs | 192 |
| Acceptance rate | ~85% |
| Repositories | 5 core products |
| Project | Role | Focus |
|---|---|---|
| flowx | Major driver | Platform stability, integration reliability, test coverage, security hardening, CI/Docker |
| fsaiosapi | Major driver (backend) | Request/schema validation, API security constraints, auth and token flows |
| fsaiosapp | Major contributor | API-doc proxy, secure gateway, auth alignment with backend |
| ai-flow | Major contributor | Validation middleware, Swagger/config hardening, dependency and container hygiene |
| langfuse-fsai | Major contributor | Iframe auth compliance, test stability, input validation |
- Reliability: Fixes and refactors across integration-heavy systems to reduce production risk and support load.
- Security: Dependency and image remediation, input validation, and auth/token consistency across services.
- Quality: Expanded test coverage and validation patterns so releases are safer and rollbacks rarer.
- Delivery: CI/workflow and pipeline updates for more confident shipping.
PR stats above were gathered with scripts/github-pr-stats-gh.sh (gh CLI + jq). Commit metrics are from local workspace analysis.
Side projects I build and maintain — mostly Next.js, React, and TypeScript, with a focus on clear specs and offline-first or self-hosted use.
| Project | Description |
|---|---|
| cc-feedback | In-product feedback capture: overlay UI, SDK (core + web), and API contract for issue and capability reports with optional text, voice, and screenshot. Monorepo with React overlay, multipart submit, and mock API for local dev. |
| cc-mobile-practice | Mobile web app for PDF-based exams: tap-to-place answer markers, assign tokens (A–E), import a gabarito (answer key), and review/grade attempts. Offline-first, no account; IndexedDB + react-pdf. Integrates cc-feedback for in-app feedback. |
| idle-page | Desktop-first ambient app: rotating media (images, GIFs, videos, quote cards) plus a TODO checklist from a local markdown file. Darkroom visual identity (red safe-light, film grain, vignette). File watcher, prefetch, optional Docker; settings in localStorage. |
| speed-reading | Mobile-first RSVP (Rapid Serial Visual Presentation) speed-reading web app. Next.js App Router, Web Worker tokenizer, calibration and onboarding flows. Vitest + Playwright; spec-driven implementation. |
Languages & runtime
Frontend
Backend & APIs
Data & auth
Tooling & DevOps
I use AI coding agents as a force multiplier in my daily work — not to replace thinking, but to extend it with clear instructions and guardrails.
| What I do | Why it matters |
|---|---|
| Structure tasks in phases | I break work into numbered steps (e.g. "0a. gather data; 1. produce report") so the agent has a clear sequence and I can refine one phase without redoing everything. |
| Ask for evidence-based outputs | I prefer results grounded in the codebase (commits, dependencies, lockfiles) so reports, tech stacks, and impact summaries are defensible for reviews and leadership. |
| Inject context via skills and plans | I attach skills (e.g. OS context, conventions) and reference plans or specs so the agent follows my environment and standards instead of guessing. |
| Choose the right tool for the job | I explicitly say when to use the browser, the shell, or the repo (e.g. "use the browser" for live UX analysis) so outputs match the task. |
| Iterate until value is clear | I don't stop at the first draft: I ask "does this make my contribution clear?" and request refinements (ownership matrix, noise filtering, value-by-project) until the artifact is usable. |
| Reuse one evidence chain | The same data (e.g. commit analysis) feeds contribution reports, tech stack lists, LinkedIn title, and this README — one source of truth, consistent story. |
| Design then implement | For bigger efforts I use the agent to brainstorm and document a plan (with options and trade-offs), then hand off the plan for implementation with todos and tests (including "how do I test in isolation?"). |
I also create skills and rules (e.g. OS-context discovery, browser-priority rules, project conventions) so agents behave consistently across sessions and workspaces. To stay current, I follow AI/tech trends mainly on X (Twitter) — I was an early adopter of Ralph loops and similar agentic patterns, and I keep an eye on emerging coding-agent workflows so I can evaluate and then adopt and adapt quickly.
In short: I treat agents as reliable teammates — I specify scope, provide context, ask for evidence and iteration, and reuse outputs across docs and automation so my profile and deliverables stay aligned with what I actually ship.
Full history: youtube-watch-history.md
Last synced: 05/03/2026, 08:09 (Brasília) · Next sync: ~05/03/2026, 22:00 (Brasília)
| Title | Author | |
|---|---|---|
 |
Inside America's Top Hackathon | Soon |
 |
🚨🚨 PEWDIEPIE DOES IT AGAIN... -- Then Standup -- Casey Presents NotePad++ 🚨🚨 | The PrimeTime |
 |
Making a ridiculously heavy tungsten suit (it broke my body) | NileBlue |
 |
The World's Best Watch Is Only $20 | Hunter Wears Things |
 |
A IA Quebrou o Modelo do GitHub | Augusto Galego |
- Email: jamdeveloper0@gmail.com
- Phone: +55 11 94700-7927
- LinkedIn: Juan Garcia