Iago Frota
Senior SDET · Software Engineer
Tests your team trusts, and AI you can measure.
Java and Playwright automation for critical billing, fraud-prevention and tax-document systems. AI agent evaluation and mentoring for QAs moving into SDET roles — for teams in Brazil, the US and Europe.
What I have built in production
Each case covers the problem, how I approached it and the outcome — within the limits of what I can share about a client system.
- Featured case
Quality on a tax-document platform: coverage that found real defects
The suite found four real defects in a regulated domain: a null pointer in the callback, an unmapped payment combination, a non-deterministic query and a type-conversion error.
Read the case (in Portuguese) → -
Test layer for a fraud-detection engine
Automated scenarios for every decision rule, which surfaced ambiguities in the rules themselves. The domain knowledge became a service contribution: the restriction-list adapter.
Read the case (in Portuguese) → -
PoC: feasibility of moving a legacy regression suite from Shell to Java
A feasibility study on a suite with over a decade of Shell: reverse engineering, two PoCs and the design of the equivalence check. The migration itself has not moved forward yet.
Read the case (in Portuguese) → -
Billing regression automation (Oracle BRM)
Declarative test plans asserting on accounting entries, centralized business rules and unit tests for the test tooling itself.
Read the case (in Portuguese) → -
Comparative PoC: Selenium + Java vs. Cypress
A comparison run against the actual Angular SPA, with explicit criteria. Cypress was argued for and adopted as the UI suite standard.
Read the case (in Portuguese) →
Measured, not estimated
- 195AI requests measured across 24 sessions, with no productivity gain to claim
- 916content pages in an LLM-maintained wiki, measured up to where it breaks
- 219pages in that wiki linking three or more sources
Recent articles
Testing, quality, career and AI applied to engineering — always from what I have actually used.
-
Escala 5x2: o projeto pessoal que nasceu de um áudio do meu irmão
Um áudio no WhatsApp virou uma PoC de escala 5x2 no ar em uma noite e, agora, um sistema para lojas e contadores, com Laravel, PostgreSQL e agentes de IA.
-
LLM Wiki na prática: o que eu apresentei no Pitch Tech da NTConsult
A apresentação de 12 de novembro por escrito: a noite em que a wiki se pagou, o fluxo, os números, quatro erros reais e um kit mínimo para você montar a sua.
-
Brag Document: como venho documentando minhas entregas e o agente que estou criando
Minha experiência registrando entregas, o relatório pedido às 19h21 para a manhã seguinte e o agente de Brag Document que estou criando para o Anvilore.
-
Estudando para a certificação de developer da Anthropic: descobri que o que eu já fazia tinha nome
Estudo 1h por dia para a certificação de developer da Anthropic e descobri que compactação, anotações e subagents que eu já usava têm nome.
-
Git Flow: o que aprendi quando os branches dos times se misturaram
Um caso real de branches misturados, os problemas ao subir para a release e um exemplo de Git Flow, com seus benefícios e limites.
Follow the next articles
Subscribe to the RSS feed or follow me on LinkedIn, where I post when a new piece is out.
Nine years shipping software, four of them as a Senior SDET.
I work with engineering teams that need their test suite to mean something: frameworks people trust, pipelines that don’t lie and quality processes that survive a sprint.
I write Java and PHP, automate browsers and APIs with Playwright and TypeScript, and mentor QAs moving into SDET roles in the international market.
I build and evaluate AI systems: versioned context for coding agents, LLM and agent evaluation, and AI-generated tests measured against real data. I study at least one hour a day for Anthropic’s developer certification, still in progress.
Stack
Let’s talk about the role — or about your test suite.
Book a 20-minute call. I’ll tell you plainly whether I’m the right person for what you’re building.