Leandro Marcos Da Silva Email and Phone Number

Fraud detection in financial transactions applying machine learning.

Development of an anomaly-based Intrusion Detection System (IDS) for swarms of Unmanned Aerial Vehicles (UAVs); Analysis of flight samples and selection of physical and geographic features to provide hardware independence; Application of supervised and unsupervised machine learning techniques to identify attacks on the UAV network and flight anomalies, respectively; Employing federated learning to ensure flight data privacy and training collaboration; Implementation of IDS in vehicles to ensure security by alerting the Ground Control Station (GCS) when an abnormality occurs, in addition to low energy consumption and use of hardware resources.Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Scholarship: 04/2022 - 10/2023.

Teaching Improvement Program (PAE) internship in the subject "Computer Network" in the Institute of Mathematics and Computer Sciences at the University of Sao Paulo (USP).

Checking students' tasks by providing constructive feedback; Updating spreadsheets with student progress; Conducting mentoring sessions via call; Answering student questions in chats; Teaching classes on basic to advanced Python, SQL, Git and GitHub, front-end technologies, and Django.

Collection of simulated flight data on Multi-Robot Systems (MRS); Testing attacks directly on the Gazebo GPS plugin, after modifying the "gazebo_gps_plugin.cpp" and "gazebo_gps_plugin.h" files and recompiling, and injecting failures with the MAVSDK failures plugin; Detecting anomalies in UAV swarm sensors, such as GPS spoofing and jamming, using a stacked autoencoder with federated learning.

Teaching Improvement Program (PAE) internship in the subject "Distributed Computing Systems" in the Institute of Mathematics and Computer Sciences at the University of Sao Paulo (USP)

Project funded by Brazilian Network Information Center (NIC.br) to classify newly registered domains through passive DNS, which applies five models to monitor the first month of domain life after the first query, with each model referring to a specific time interval; Collection of data from the authoritative server of a Top-Level Domain (TLD), requiring the use of Big Data tools, such as Apache Spark and Hadoop; Extraction of 20 DNS features with geolocation information; Combination of Cluster Centroids and K-Means SMOTE for data balancing; Comparison of five tree-based algorithms, including XGBoost and LightGBM; Testing models built with new data collected from the TLD. In short, the project covered the following steps:· Filtering data from only newly registered domains;· Selection of data based on a time interval;· Data pre-processing;· Feature extraction and data labeling;· Data balancing;· Training and validation of each model;· Model testing.