Skip to content
.dev

Flavio Gouveia

Senior Frontend & Mobile Engineer

I work on TED.com and the TED app in React Native, products that hold up in production every day. I use Claude Code and Codex on real code, with specs, tests and human review deciding what ships.

Available for freelance workEsposende, Portugal · remote, EU/US time zones

Discuss a projectView resume
Current work: TED Conferences2023 to present

  • 20M+monthly visitors on TED.com

  • 30M+TED app downloads on Android

  • 3,000+TED talks with subtitles · 100+ subtitle languages

  • 5+years remote with US teams

  • WCAG 2.1 AA on core page templates · Playwright and Jest in CI blocking regressions

9+ years building web and mobile

Where I've shipped

Open any commit to see the diff of what shipped.

  1. currentApr 2023 to present

    • Front-end development of TED.com, a platform serving 20M+ monthly visitors, by architecting page rendering and content delivery in Next.js, React, GraphQL, and Tailwind CSS.

    • Sustained performance and stability on a React Native app with 30M+ downloads on Android alone, by monitoring production health and release stability in New Relic and prioritizing fixes by user impact.

  2. Aug 2021 to Apr 2023

    • Shipped RadStar Alpha Series to the App Store and Google Play for Spruce Environmental Technologies, a tablet-first field application for professional radon testers, by building — from a single React Native codebase for iOS and Android — the workflows that control continuous radon monitors and sync test results, reporting templates, and quality-control records to a cloud platform.

    • Shipped Melanin Medical Specialties to the App Store and Google Play, a cross-platform React Native app connecting healthcare professionals and students, by delivering mentor matching, a searchable community provider directory, and peer-support features.

  3. Sep 2019 to Aug 2021

    • Enabled coordinated maintenance shutdowns across hydroelectric, thermoelectric, and wind power plants operated by Engie nationwide, by building the React operational interfaces and Node.js/TypeScript APIs plant operators depend on.

    • Delivered the full stack of a system where downtime carried direct revenue and regulatory consequences, by hardening API error handling and optimizing high-volume asset queries on Oracle Database.

  4. Jun 2018 to Sep 2019

    • Led a team of 8 developers across back-end, front-end, and React Native mobile, by setting coding standards, running code review, and defining delivery cadence.

    • Shipped loyalty and engagement platforms to the Brazilian market, by architecting a React web client, a React Native mobile app, and Node.js services on a shared PostgreSQL data model.

  5. May 2017 to Jun 2018

    • Full-Stack Developer at SESI (Oct 2017 – Jun 2018): built a wellness coaching application delivering personalized health programs to industrial workers, using a Java back-end and Angular front-end on MySQL.

    • Software Development Intern at SumOne (May 2017 – Oct 2017): contributed to client web applications in JavaScript while completing an Information Systems degree.

  6. B.Sc. Information Systems

    Universidade Federal de Santa Catarina (UFSC), Brazil

    initial commit2016 – 2020

Open-source AI projects

Public code on GitHub: agents, MCP and RAG, with security checks and evals in CI.

  1. feature/incident-copilot

    Multi-agent incident copilot in TypeScript: a LangGraph supervisor coordinates telemetry, runbook, planning and audit agents; risky actions go through human approval, and forbidden ones never run even when the model proposes them.

    Tech stack: TypeScriptLangGraphMCPReactNode.js

    View on GitHub: incident-copilot (opens in a new tab)

  2. feature/secure-legacy-mcp

    A stdio MCP server that turns a hostile legacy REST API into business actions behind a real security layer: hashed, revocable service tokens, server-side RBAC and rate limits.

    Tech stack: TypeScriptMCPRBACNode.js

    View on GitHub: secure-legacy-mcp (opens in a new tab)

  3. feature/docs-data-assistant

    A documents-and-data assistant that refuses when evidence is weak, cites its sources, and runs read-only validated Text-to-SQL, with guardrails and an eval gate in CI.

    Tech stack: TypeScriptRAGText-to-SQLEvals

    View on GitHub: docs-data-assistant (opens in a new tab)

How I work with AI

AI speeds up delivery. The spec, the tests and the review decide what reaches production.

spec,failing tests,human review

  1. Spec before code

    The agent gets a spec and failing tests before it writes a single line of implementation.

  2. Tests are the gate

    Playwright and Jest in CI decide what ships, not the model's confidence.

  3. Human review, every time

    Every AI-generated diff goes through code review, and agent permissions stay scoped.

Merge only when all three gates pass.

Have a web or mobile project?

Freelance work in frontend, React Native and applied AI. Tell me what you need.

Or reach me directly

Project type