• Trained state-of-the-art U-Net-based models for brain tumor segmentation and brain extraction (brain segmentation)• Authored a journal article on deep-learning-based brain extraction methods in the tumor segmentation pipeline• Authored a conference paper on pre-training deep learning models for brain age estimation• Currently working on post-hoc confidence estimation for selective prediction on image segmentation tasks

• Developed and implemented an anomaly detector for time series data from an industry partner;• Co-authored a scientific publication on automatic machine learning application in the production environment;• Implemented and automated a data preprocessing pipeline for a turbomachinery milling process (+250 GB per dataset).

Main activities:• Lead 15 high qualified engineering students, managing their performances and interests;• Study, organize and implement the group's strategic plan;• Manage the expectation of the 4 partner companies: WEG, Embraco, CERTI Foundation and BIX Technology;

NEO Empresarial is a program designed to develop engineering students with technical, managerial and personal abilities through its self administration as well as the technical execution of projects in partnerships with industrial and technological companies (Embraco, CERTI Foundation, BIX Technology and WEG).My main achievements at NEO:• Implemented an ARIMA and a LSTM Network using Python for time series forecasting, aiming to reduce the prediction-making process and to improve its accuracy;• Developed a library in C for micro-controllers to communicate with an I2C MEMs sensor, for a specific application;• Developed a software to manage tests of an industrial laboratory, providing the users the ability to visualize and compare multiple results, improving the performance analysis of the tested materials.

Automatization of a bearing fault detection test using envelope analysis through LabVIEWsoftware development.• Multiple accelerometers data acquisition using National Instruments DAQmxlibrary;• Acceleration signal processing to generate envelope signal and its magnitudespectrum, besides other auxiliary analyis;• Software validation using large motors.

Development of a new production planning and control system for a disassembly line, bythe adaptation of main concepts of PPC for assembly lines. Additionally, a new demandprevisibility system was developed using statistical approaches to avoid future fluctuationsin the production.• Analysis and adaptation of PCP and reverse manufacturing related theory;• Enhancement of production previsibility improving human resources’ efficiencythrough idleness reduction.;