Engineer first. Consultant second.
Senior Data & AI Engineer. 28 years shipping software, ~15 of those specialized in data platforms, the last two pulling AI into production data work. Rio de Janeiro, GMT-3.
The short version
I design and operate end-to-end data systems — ingestion, modeling, serving, observability — from mid-market SaaS to petabyte ad tech. Over the last two years I've shipped Anthropic Claude agents (guardrails, skills, tool-use orchestration, MCP) into day-to-day data engineering workflows. I write here about what actually holds up in production and what doesn't.
The path
Six global-scale companies, a few smaller ones, and one long thread: making data systems reliable enough that other people can rely on them.
Major B2B Data Intelligence Platform — Senior Data Engineer (2024–2026)
Sole data engineer embedded in Sales & Marketing. Rebuilt the finance pipeline: replaced a recurring monthly manual data-fix cycle with an automated detect-and-remediate system that runs unattended at close. Introduced Claude agents with guardrails and tool-use into development and maintenance workflows. Extended with Model Context Protocol (MCP) in early 2026.
Fast-Growing Logistics Startup — Tech Leader / Senior Software Developer (2023–2024)
Stood up the entire data function from zero. Designed the governance architecture — lake, warehouse, analytics — on Google Cloud. Built ingestion pipelines from four heterogeneous source systems into BigQuery. Defined the team's charter and initial delivery roadmap so the internal team could own the platform going forward.
Global Domains & Hosting Platform — Senior Data Software Engineer (2022–2023)
M&A integration engineering. Absorbed pipelines and services from acquired companies into the parent platform. Brought acquired systems into GDPR compliance (PII handling, consent, retention). Migrated services from AWS to GCP.
Leading Global Travel Platform — Senior Software Engineer (2021–2022)
Built the new data pipeline on AWS (S3, EMR, Glue, Redshift, Lambda, DynamoDB). Created the data lake that unblocked the ML team — they could work with near real-time company data without impacting production.
Top-10 Global Ad Tech Platform — Data Software Engineer (2017–2021)
Data team at one of the world's largest ad platforms. 1.5 billion records per day at petabyte scale. AWS EMR/Hadoop, Scala/Java/Python. Powered real-time retargeting, prospecting, and email optimization for thousands of advertisers. This is where I learned how big data actually behaves under load.
Earlier stops
A leading Brazilian e-commerce (tech lead, hackathon that cleared ~70% of the small-bug backlog and killed a recurring 6-hour Saturday outage), a Brazilian fintech payments company, a digital agency, Brazil's largest media company (5+ years), enterprise consulting, a European telecom software vendor in Germany, a Brazilian math research institute, and my first job — trainee at a major telecom operator in 1998.
Why data and AI together
Data engineering is where the inputs to every AI system come from. If the inputs are wrong, the model output is wrong — and nobody notices until someone above you reads a dashboard and asks a question nobody can answer. LLMs change what a data team can do inside its own workflows: review code they'd never review before, catch regressions earlier, document what they build, answer questions about pipelines without rebuilding tribal knowledge. That's the integration I've been shipping and writing about.
How I work
Short feedback loops. One change at a time. Tests before trust. Observability as a first-class citizen, not an afterthought. I prefer the boring, obvious solution unless there's a real reason to be clever. I read a lot and write as a way of thinking — this site is part of that habit.
Contact
Best reached via LinkedIn or GitHub. For consulting engagements, see mentgesinformatica.com.br .