Distributed Systems Engineer
I'm a Distributed Systems Engineer with a focus on smart city technologies. Originally from São Paulo, Brazil, I'm now based in France, where I am pursuing a Masters in Computer Engineering at Sorbonne Université. My passion lies in developing solutions for urban challenges using software and hardware technologies.
Sorbonne Université
Université Paris Cité (Paris Descartes & Diderot)
Virtualization Engineer Intern (March 2025 - September 2025)
Solutions Architect Intern (Summer 2024)
Software Engineer Intern (Summer 2022)
Developed a comprehensive digital twin system for urban infrastructure monitoring and disaster risk management. This AWS-based solution integrates IoT sensors, real-time data processing, and AI-powered risk assessment to enhance city safety and infrastructure maintenance. Key features include a 3D visualization of city infrastructure, real-time sensor data analysis, and an alert system for potential hazards. The project utilizes various AWS services such as IoT Core, Lambda, SQS, ECS, and Athena, along with custom hardware solutions for data collection.
Developed a system to improve meteorological data quality for flood and drought prevention. This project, part of the UN-backed Crowd4SDG program, aims to enhance the reliability of weather forecasts by optimizing data flow between collectors and hydrological models. Potamoi implements real-time data cleaning techniques to externalize rapid quality control, reducing costs and improving forecast accuracy without requiring additional internal expertise. The project was presented to UN experts, highlighting its potential impact on climate change mitigation and urban problem-solving through technology.
Developed a machine learning model for recognizing fire hazards in surveillance camera footage and environment categorization. This project uses Convolutional Neural Networks (CNNs) to process images and classify them for potential fire risks. The system includes multiple models for tasks such as fire detection, object classification (CIFAR10), pedestrian detection, and scene classification. Implemented as a web application, it allows users to upload images for analysis, demonstrating the potential for real-time fire hazard detection in smart city environments. The project aims to improve emergency response times and enhance urban safety through automated surveillance processing.
Developed a solution for multi-process memory-mapped file sharing that ensures data integrity and isolation. This project implements a copy-on-write mechanism to allow multiple processes to read from a shared memory-mapped file while isolating write operations. Key features include signal handling for write attempts, memory page duplication, change logging, and a merge system for consolidating modifications. The solution utilizes techniques such as page table manipulation via the PTEditor API and custom signal handlers. This approach improves upon traditional file I/O methods by reducing system call overhead and providing efficient, conflict-free concurrent access to shared files in a multi-process environment.
Enhanced the Ouichefs file system for Linux 6.5.7 to optimize data insertion in the middle of files. This modification allows for more flexible block filling, significantly improving efficiency when editing large files. The project includes reimplemented read/write functions, smart data organization, and file defragmentation features.
Developed an emergency response system for public parks, integrating hardware and software components to ensure visitor safety. The system features strategically placed Emergency Alert Stations with multiple buttons for different emergency types, LED indicators for status updates, and a Central Monitoring System for real-time alert tracking and response coordination. Key components include a Java-based backend using Spring Boot, a web interface for park staff, and potential integration with local authorities.
Developed a client-server communication system implementing a "Megaphone like" protocol. This project features a multi-threaded server capable of handling multiple client connections simultaneously, supporting both IPv4 and IPv6. Key functionalities include user registration, creating and posting to discussion threads, subscribing to threads for notifications, and file sharing capabilities. The system utilizes TCP for reliable communication and UDP for efficient file transfers. Advanced features include real-time notifications via multicast, thread subscriptions, and robust error handling.
Developed a park sensor network system for real-time environmental monitoring. The project integrates a Java Swing application for user interaction, a backend server for data processing, an IoT sensor network using ESP32 boards, and a cloud-hosted PostgreSQL database. Key features include a user-friendly interface displaying park maps with sensor locations, real-time data visualization from various environmental sensors (air quality, water quality, soil humidity, and microclimate), and a scalable architecture for future sensor additions.
Developed a shell-like command-line interface that mimics and extends the functionality of traditional Unix shells. This project features a modular architecture with components for command interpretation, execution of both internal and external commands, wildcard expansion, and advanced I/O redirection including pipeline support. Key features include custom implementations of internal commands like 'cd', 'pwd', and 'exit', along with the ability to execute external commands.
Graphical application for pathfinding algorithms
Managed application security process for internal AWS web application
Exploratory Data Analysis and Machine Learning for urban traffic behavior
Developed a smart waste management system aimed at optimizing garbage collection in Paris, France. The project integrates IoT devices, data analytics, and route optimization to improve urban waste management. Key features include real-time monitoring of trash bin fullness using 3D modeling and volume calculation, an intuitive interface for waste management teams, and an advanced algorithm for determining optimal garbage collection routes. The system employs Arduino-based sensors in its initial phase.
Practice project using Sklearn Breast Cancer dataset for classification