Gustavo Melo

Data Engineer · Brasília, DF

I build pipelines that keep messy data honest.

Software Engineering student focused on data engineering. I design ingestion, modeling, and data-quality systems that run in production, not just in notebooks, currently as part of the data team behind GovHub at LabLivre.

01 · About

Where the data comes from matters

I'm a Software Engineering student in Brasília, Brazil, in my final year, focused on data engineering. Most of what I know comes from building things that actually run in production: pipelines that move large volumes of files without losing any of them, checks that catch bad data before it reaches a dashboard, and systems that keep working long after I've moved on to the next problem.

I currently work on the data team behind GovHub, a LabLivre project that consolidates municipal datasets from Brazil's Ministry of Cities into a single data lake. Before that, I worked on SciNewsAI, an academic project that uses retrieval-augmented generation to turn scientific papers into accessible content.

I care about pipelines that are boring in the best way: predictable, well documented, and easy for the next engineer to pick up.

Based in
Brasília, DF (UTC-3)
Currently
Data Engineering Intern @ LabLivre / GovHub
Studying
Software Engineering, UnB
Open to
Remote data engineering roles

02 · Experience

Experience

Data Engineering Intern — LabLivre · GovHub Project FEB 2026 – PRESENT · BRASÍLIA, DF

Part of the data team behind GovHub, consolidating municipal datasets from different government agencies into a unified, LGPD-compliant data lake.

  • Implemented the dbt pipeline (bronze → silver → gold) for Minha Casa, Minha Vida (FAR) housing developments, modeling construction progress for a national map and state-level dashboards.
  • Built ingestion DAGs in Apache Airflow to bring municipal datasets from SFTP into a data lake on MinIO, with file-hash idempotency and parallel multipart upload for files up to 37 GB.
  • Built API and web scraping ingestion DAGs for FGV economic indicators, replicating the authentication flow of a legacy OutSystems/ASP.NET system to automate a manual process.
  • Implemented deterministic HMAC-SHA256 masking for sensitive data (CPF, ZIP code, address) across 1,600+ files, preserving join integrity across datasets.
  • Designed the lake's bronze layer in typed Parquet, replacing ~2,000 untyped VARCHAR tables with a single incremental, idempotent architecture.
Python Airflow dbt MinIO (S3) PostgreSQL DuckDB
Data & Pipeline Contributor — SciNewsAI · UnB AUG 2025 – DEC 2025 · ACADEMIC PROJECT

Platform that monitors arXiv papers and uses generative AI/RAG to turn them into accessible content via blog and newsletter.

  • Built automated collection pipelines (scraping and APIs) and metadata preprocessing to feed the paper catalog.
  • Integrated retrieval and generation components and orchestrated the flow with Docker across scrapers, the AI pipeline, and web services.
Python Pandas Selenium Docker

03 · Projects

Selected work

Data lake · Production

Municipal Data Lake — GovHub

End-to-end ingestion for hundreds of municipal data sources: SFTP pipelines with file-hash idempotency, a masking layer for personal data, and a typed Parquet bronze layer that replaced ~2,000 untyped tables with a single incremental architecture.

Airflow dbt DuckDB

Web scraping · Reverse engineering

Automating a Legacy Login Flow

FGV publishes economic indicators behind a legacy OutSystems/ASP.NET login with no public API. I reproduced its authentication flow (dynamic hashes, ViewState, DevExpress grid) in Python to automate a collection that used to be done by hand every month.

Python Requests Airflow

RAG · Academic project

SciNewsAI

A platform that monitors arXiv, ranks papers by relevance, and uses retrieval-augmented generation to turn dense academic writing into accessible summaries for a blog and newsletter.

Python RAG Docker

04 · Skills

Toolbox

Languages & data

Python
SQL
Pandas

Orchestration & modeling

Apache Airflow
dbt
DuckDB

Storage & ingestion

PostgreSQL
MinIO (S3)
REST APIs
Web Scraping

Quality & ops

Data masking (LGPD)
Idempotent pipelines
Docker
Git
Linux

05 · Education

Education

Software Engineering

UnB – University of Brasília
July 2021 – Present

Advanced English

FISK – Fundação Richard Hugh Fisk
January 2014 – December 2018

06 · Contact

Let's build something reliable.

Open to remote data engineering roles and internships. The fastest way to reach me is email.

or call +55 (61) 98289-6001 · message on WhatsApp