I like understanding how things work and using data to solve problems that make a difference in people’s lives.
Here I share projects, questions and what I learn by trying things out. Whenever I can, I put the code and data on GitHub for you to explore too.
Data and Machine Learning · Working internationally
From Seville, Spain, to wherever you are.
A straight-line estimate using your approximate IP location from ipwho.is. VPNs and mobile networks may change the result. This feature does not store your IP or coordinates.
What I've been working on
Part 1 of the series on the Socioterritorial Conditions Index (ICST) of Maricá (RJ, Brazil). Why the municipal average does not help decide where to invest, how to turn ten official variables into a 0–100 index per census tract, and what to do when the data does not exist in the form you wanted.
Part 2 of the series on the ICST of Maricá (RJ, Brazil). The index ranges from 30.9 to 96.1 across census tracts, sanitation is the weakest dimension and disadvantages form contiguous patches. Maps, neighborhoods and districts, with a step-by-step calculation example.
Final part of the series on the ICST of Maricá (RJ, Brazil). 78 robust priority tracts, with 27,683 residents (14% of the population), concentrated in Itaipuaçu, Chácaras de Inoã and the fringes of the 2nd district. How I tested the stability of that list, why the Census shows 35% network water and SINISA 78%, and what the index does not say.
Work where I use data to investigate problems and test solutions.
I compared machine learning models with classical methods for choosing distribution centers. Here I share what worked in the experiments and what remained unanswered.
Read the analysisUsing Olist and IBGE data, I modeled where to open distribution centers and how to allocate demand. I compare forecasts, MILP and heuristics; the code and experiments are in the repository.
At RioFlex/FlexIA I was responsible for the AI agent, data and AWS infrastructure. A finalist in the ONS + UFRJ/COPPE Hackathon, the project combines document search and SQL queries for the electricity sector.
I contributed to this QGIS plugin for georeferencing historical aerial photographs. As a technical lead and reviewer, I worked on code organization, the pipeline and multithreaded processing.
I worked as technical lead and full stack developer on this geoinformation request system, from requirements through backend, frontend and technical reviews. It includes five approval levels, auditing and map sheet selection.
Languages and tools I use in my work.
My work combines data preparation, systems development and model evaluation.
I contribute to vegetation and hydrography classification with Random Forest and XGBoost, preparing features and evaluating models with spatial validation. I also work with fine-tuning and image super-resolution.
I develop agents that query tools and documents. At FlexIA I worked with official data, sourced answers and evaluation against reference questions, plus controls to restrict SQL queries to reading.
I design and maintain ETL pipelines with Python, Java and SQL. I organize data layers and work with relational models, queries and spatial indexes in PostgreSQL/PostGIS.
I use QGIS, Google Earth Engine and FME Safe to analyze territories, produce maps and validate geometries. I also develop PyQGIS plugins and integrate sources with GDAL, PostGIS and GeoPandas.
I develop APIs with FastAPI and Spring Boot, use Docker and Kubernetes for containers and automate tasks with n8n. I care about making systems easy to maintain and review.
I have worked with data for over ten years. These are some of the experiences that brought me here.
Data Engineer and Data Analyst · freelance consultant
Consulting and analysis to translate operational goals into variables, constraints, and metrics. I modeled capacitated distribution-center location for a client in another sector, combining MILP, heuristics, matheuristics, and demand forecasting. I reapplied the approach to public e-commerce data to document and validate the method.
Exército Brasileiro ·
Development of Python/SQL pipelines, PostgreSQL/PostGIS databases and cartographic automation. Contributions to Random Forest and XGBoost classification models, spatial cross-validation, fine-tuning and image super-resolution. QGIS plugin development and technical leadership in geoinformation systems.
Engprat · Contract · Rio de Janeiro, hybrid
Designed a relational schema in PostgreSQL/PostGIS from scratch to consolidate field data from thousands of properties and residents. Python ETL in raw/staging/production layers, automated spatial validations (boundary overlaps, attribute coherence), territorial density analysis via neighborhood analysis, and results exposed through a REST API consumed by external systems.
Exército Brasileiro
Design and implementation of automated validation workflows in Python and PostgreSQL. Definition of formal spatial rules, systemic topological consistency controls and structured error logging at every pipeline stage. The system enabled scaling cartographic production while maintaining verifiable technical standards and reducing rework.
COBE Engenharia e Geotecnia · Freelance · São Paulo
Processing and integration of bathymetric sensor data for the Porto Primavera Hydroelectric Plant. Digital Elevation Model generation, volume calculation per elevation and hydraulic flow analysis. Full pipeline with Python, SQL and FME Safe for conversion, transformation and standardization of heterogeneous spatial datasets.
Exército Brasileiro
Logical and physical design of PostgreSQL/PostGIS databases for complex geospatial data. Spatial query optimization with GiST/SP-GiST indexing. ETL process development with FME Safe and Python for integration of heterogeneous sources — vector files, rasters, spreadsheets and geospatial web services.
If you want to discuss a project, exchange ideas or work with me, you can reach me here.