Dublin, Ireland
Senior Machine Learning Scientist
Intercom
Doctor of Philosophy (Ph.D.) - Computer Science, Deep Learning, Speech Production
Multispeech Research Team
Loria, Inria, CNRS, Univ. de Lorraine
Thesis Title: Deep supervision of the vocal tract shape for articulatory synthesis of speech
Extra-curricular activities:
Master of Science - Electrical Engineering, Deep Learning, Computer Vision
Department of Computer Engineering and Industrial Automation (DCA)
School of Electrical and Computer Engineering (FEEC)
University of Campinas (Unicamp)
Dissertation Title: The Impact of Annotation Quality on Deep Learning for Skin Lesion Segmentation
Access here
Extra-curricular activities:
Bachelor of Science - Electrical Engineering
School of Electrical and Computer Engineering (FEEC)
University of Campinas (Unicamp)
Extra-curricular activities:
Proven professional experiences with the following technologies
At the AI Speech Team, I explored wake word detection and different phonetic alignment approaches for these models. I compared alignment-based and alignment-free, and proposed a hybrid alignment, that mixes the two. My experiments showed alignment-free training outperforms alingment-based for lower FAh levels, while the opposite happens on higher FAh levels. Moreover, when trained with the correct aligned-to-unaligned ratio, the proposed hybrid system obtains the best of the two worlds.
At Nexa Digital (acquired by DASA), I worked as a data scientist in the processing of medical documents using traditional ML and DL models (LSTM, CNN, attention mechanisms). Apart from traditional data science, my work included developing data infrastructure, building high-performance systems and pipelines.
After the M&A, I worked on processing medical images (MRI, X-ray, ultrasound) using deep convolutional networks. Our goal was to improve the clinical daily tasks and patients' well being with artificial intelligence.
At Rapiddo Delivery (acquired by iFood), I worked as a data scientist in the food delivery business. My most significant contribution was the development of a dynamic pricing system based on the supply and demand at any time. As a result, the system enabled a positive margin in the largest city that we operated for the first time.
After the M&A, I worked as a data scientist in the logistics team. I also led and launched our mPOS system, one of our most strategic projects in that year. I was responsible for dealing with suppliers, the tech team, and other stakeholders. As a result, I was the main responsible for increasing our demand by 30% in all cities we launched.
At Eldorado, I worked at an international project for Motorola Mobility. We were responsible for developing software automation tools to help the engineering teams to be more productive. As a software engineer, I worked mainly with Python at a team with people from the US, Brazil, and India.