Tiago Benetti
Data Scientist / Innovation Designer at Instramed | Cardioversors and Defibrillators- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
Espanhol Elementary proficiency
-
Português Native or bilingual proficiency
-
Inglês Professional working proficiency
Topline Score
Bio
Experience
-
Instramed Indústria Médico Hospitalar Ltda.
-
Hospitals and Health Care
-
1 - 100 Employee
-
Data Scientist / Innovation Designer
-
Aug 2019 - Present
- Research of state-of-art algorithms, methods and technologies for cardiac monitoring and cardiac therapy. Focus on biomedical signal/data such as Electrocardiogram (EEG) and Photoplethysmogram (PPG) - Data preparation of biomedical datasets, signal and data processing using classic and advanced DSP techniques (adaptive filters, spectral estimation, STFT, Wavelets), extraction and selection of features for time series biomedical data (temporal, statistical and spectral features), reduction of data dimensionality and handling imbalanced data. - Research and development of Machine Learning models for classification and regression problems of time series biomedical data. I worked with models from classifiers based on SVM, Random Forests, k-NN and Neural Networks (MLP, RNN and CNN) - Model deployment on desktop computers (C# .NET) and embedded devices (C/C++) - Tools and languages : Python (numpy, pandas, sklearn, keras, tensorflow, pytorch) Matlab, C, C++, C# .NET, ML .NET, ONNX - Monitoring of external projects with Universities and Research Institutes in Biomedical Engineering field - Development of project's specification and requirements analysis Show less
-
-
-
Borges engenharia
-
Brazil
-
Higher Education
-
700 & Above Employee
-
Embedded Software Developer
-
Sep 2018 - Mar 2019
I was busy developing a personal project aimed at implementing the heart rate estimation algorithms developed during my Masters in Embedded Linux Software (C/C++ language) for a hardware based on ARM Cortex-A8 + DSP TMS3206xxx processor. - Later, I refactored the code, simplifying the algorithm to reduce the computational cost. Successfully implemented it on an ARM Cortex-M4 STM32F4xx microcontroller. I was busy developing a personal project aimed at implementing the heart rate estimation algorithms developed during my Masters in Embedded Linux Software (C/C++ language) for a hardware based on ARM Cortex-A8 + DSP TMS3206xxx processor. - Later, I refactored the code, simplifying the algorithm to reduce the computational cost. Successfully implemented it on an ARM Cortex-M4 STM32F4xx microcontroller.
-
-
-
Centro de Pesquisa em Microgravidade da PUCRS (MicroG)
-
Porto Alegre, Rio Grande do Sul, Brasil
-
Signal Processing Researcher and Developer
-
Mar 2016 - Aug 2018
- In the first months, I was involved with research related to hypoxia in aerospace environments and in Microgravity. I participated in the acquisition of signals through the development of hardware prototypes for the acquisition of biomedical signals (ECG, PPG) in the laboratory's hypobaric chamber. - Later, I joined the academic master's degree in electrical engineering at PUCRS, where I was involved in researching the effects of Microgravity on human physical activity and chose the line of research as obtaining heart rate in pulse oximetry devices. - Research and Development of methods for estimating heart rate in smart watches (smartwatches) from pulse photoplethysmography (PPG) signals of individuals performing physical activity - The algorithm for estimating heart rate, developed in Matlab, used adaptive filters (NLMS, RLS algorithms) in the adaptive noise cancellation approach with the acceleration signal as a reference to attenuate the motion artifacts in the signal. The Welch periodogram in the signal window provided an initial estimate of heart rate and the Kalman filter adjusted with heuristics based on signal quality tracked the heart rate in real time. A signal quality index was developed with features and parametric estimation methods and later, through the categorization of signal quality with the K-means algorithm. The methods were validated in a public database of individuals practicing physical activity. * Scholarship by CAPES and HP Show less
-
-
-
LABIMA PUCRS
-
Porto Alegre, Rio Grande do Sul, Brasil
-
Image Processing Developer
-
Jan 2015 - Dec 2015
- Research and Development of methods for separating and removing shadows from images in large urban areas. - Implemented in Matlab the Otsu's method, which is based on an automatic threshold applied to the image's histogram. The threshold was chosen by Linear Discriminant Analysis in order to seek maximum separation between shadows and other elements of the image. - I also implemented in Matlab the method for shadow separation based on the K-means algorithm - Researched and developed improvements to the algorithms Show less
-
-
-
Datacom
-
Brazil
-
Telecommunications
-
300 - 400 Employee
-
Embedded Software Engineer
-
Oct 2010 - Mar 2013
- Development of C language software for Embedded Linux in PowerPC microprocessors for Metro Ethernet Switches. - Development of embedded software involving specification, implementation and testing of L2 layer protocols. - Implementation of board support packages (BSPs) for hardware platforms: development and maintenance of Bootloader (u-boot), Linux Kernel and Device Drivers for new and old products. - Support and maintenance of the Software Development Kit (SDK) of Ethernet Switch ICs at the application layer. - Development of software for managing hardware devices at the application layer. - Bug fixes and software optimization. - Development of Expect and Python scripts for testing and validation of software. - Version control with Git and CVS. - Experience with Agile Project Management with Scrum - Responsible for the integration of software version and the analysis of results obtained in automatic tests during a period of six months Show less
-
-
-
Laboratório de Instrumentação Eletro-eletrônica - UFRGS
-
Porto Alegre, Rio Grande do Sul, Brasil
-
Graduate researcher
-
Oct 2009 - Nov 2010
- Research and development of methods to recognize hand movement patterns in Electroencephalography (EEG) signals using Machine Learning for Brain-Computer Interfaces - Developed a Matlab model for a movement classifier using a Multilayer Perceptron Neural Network (MLP). The model had as input features extracted from a range of the 6-channel EEG spectrum of the motor cortex and was based on the paradigm of motor imagination. The features were obtained by decomposing the signal using the Discrete Wavelet transform. Dimensionality reduction was performed with Principal Component Analysis (PCA). - Scholarship by CAPES for an academic master's degree in Electrical Engineering Show less
-
-
-
LABIMA PUCRS
-
Porto Alegre Area, Brazil
-
Software Developer
-
Aug 2007 - Jul 2009
- Firmware and Software Developer in the project of a Graphic Tactile Monitor to assist the visually impaired people. - Development of software for Windows environment in C ++ language. The software aimed at the digital capture and processing of screen images of personal computers and the transmission of data to an embedded device. Technologies involved: Visual Studio, OpenCV, RFB protocol, USB protocol. - Development of firmware in C language for embedded device based on MSP430 microcontrollers, aiming the reception of data and the activation of an electromechanical system. Technologies involved: IAR Embedded Workbench, I2C Protocol. Show less
-
-
-
AES Sul Distribuidora Gaucha de Energia SA
-
1 - 100 Employee
-
Research intern
-
Apr 2004 - Mar 2005
- CAD design and simulation of a helical hydraulic turbine for small hydropower in computational fluid dynamics software tool. The project complied with a program to reduce the waste of electricity in Energy Management with a focus on the supply side. - Involved Tools: ANSYS CFX - Prize of prominence in the V Hall of Scientific Initiation in Engineering area of PUCRS with the work "Management of Energy: Approach on the side of Supply. Numerical Simulation of a Hydropower Turbine". - CAD design and simulation of a helical hydraulic turbine for small hydropower in computational fluid dynamics software tool. The project complied with a program to reduce the waste of electricity in Energy Management with a focus on the supply side. - Involved Tools: ANSYS CFX - Prize of prominence in the V Hall of Scientific Initiation in Engineering area of PUCRS with the work "Management of Energy: Approach on the side of Supply. Numerical Simulation of a Hydropower Turbine".
-
-
Education
-
PUCRS - Pontifícia Universidade Católica do Rio Grande do Sul
Master of Science Degree, Electrical Engineering (Digital Signal Processing, Biomedical Engineering) -
Universidade Federal do Rio Grande do Sul
Master of Science, Electrical Engineering -
Pontifícia Universidade Católica do Rio Grande do Sul
Bachelor's Degree, Mechatronics, Robotics and Control and Automation Engineering