Projects
Real projects in data engineering, geoinformation and automation — from ETL pipelines and PostGIS databases to n8n workflows and QGIS tools. Each system prioritizes traceability, data quality and maintainability.
I developed a project for locating distribution centers and assigning demand under capacity constraints, using real Olist and IBGE data. I combine demand prediction, MILP models, and heuristics to compare costs, coverage, and scenarios. On the blog, I share the algorithms, experiments, and what we can learn from their results.
AI agent and data lake for the power sector. Responsible for the agent, data and AWS infrastructure in RioFlex/FlexIA, a finalist in the ONS + UFRJ/COPPE Hackathon. RAG, hybrid search, Nemotron/Claude routing and SQL queries with guardrails.
QGIS plugin for georeferencing historical aerial photographs using RootSIFT, FLANN and RANSAC. I contributed as a tech lead and code reviewer, refining and standardizing the code according to SOLID and Clean Code principles. I also refined the pipeline and improved processing with multithreading.
Geoinformation request system with five approval levels, auditing and map-based sheet selection. I worked as a tech lead and full-stack developer, covering requirements, technical decisions, backend and frontend development, documentation and technical design reviews.
Scenario simulation through 2035 to assess how electric vehicle growth and charging times may affect grid demand. Compares unmanaged charging, smart redistribution over 24 hours and V2L, with historical peak analysis, adjustable parameters and traceable data.
A system for integrating and analyzing geospatial data on land use and public policy. Consolidates vector layers, territorial indicators and geometries in PostgreSQL/PostGIS, with QGIS visualization and API access. Supports territorial assessment and management decisions.
An application for managing operational capacity, scheduling, production and team indicators. Turns work records into auditable metrics with full traceability. A FastAPI + PostgreSQL backend with internal ETL, exposed as a REST API for other systems.
A set of PyQGIS plugins and routines for georeferencing, topological validation and cartographic production control. Automates spatial rule checks, feature extraction and quality workflows in production environments, using PostGIS as the backend.
An n8n automation lab: integration workflows across APIs, webhooks, notifications and data processing. Includes online course sales workflows, payment management, automated emails and CRM integration. Its infrastructure is separate from the static website.
Processing bathymetric sensor data for the Porto Primavera Hydroelectric Plant. Digital Elevation Model generation, volume calculation by elevation and water-flow analysis. A Python + FME Safe + PostGIS pipeline with automated spatial validation.
A community land registration and documentation system: large-scale collection of property, resident and structural data. Relational schema design in PostgreSQL/PostGIS from scratch, ETL across raw/staging/production layers, spatial validation and results exposed through a REST API.
A system for integrating and analyzing point-of-sale data: transaction consolidation, cleaning, validation and loading into PostgreSQL. Automated operational indicators with Python and n8n workflows for alerts and periodic reporting.
A custom web analytics system. A JavaScript beacon sends events to a FastAPI API that stores them in PostgreSQL and aggregates hourly metrics. A private JWT-protected dashboard runs on a separate subdomain, alongside a public page with live metrics.