Pedro Vitor Quinta de Castro

This research project consists on crawling the internet for scientific articles belonging to the medical area and making them available to a search engine, after applying a series of Natural Language Processing techniques for Information Extraction. The articles are also supposed to go through a clustering pipeline in order to improve their relevance to the end user.

Worked at EasySearch, which is an Enterprise Search Engine based on Apache Solr, Java, MongoDB and Angular 2 with Bootstrap. Introduced Artificial Intelligence and Machine Learning concepts, in order to provide more intelligence to the search. To do this I developed componentes based on Latent Dirichlet Allocation, to provide topic modeling and automatic classification of the indexed documents, as well as another based on doc2vec, to provide semantic similarity between documents. Used Python with Gensim.

Worked as an architect and technical leader for major banking institutions customers. The project for Banco do Nordeste do Brasil (BNB) was a system for financial control of legal expenses, based on Java EE technology, built around IBM WAS and RSA. Afterwards I was reassigned to the Banco Central do Brasil (BACEN) customer, where I worked with TDD and CI with Jenkins, being in charge of ensuring the applications' acceptance, by enforcing the code quality criteria by SonarQube integration.

At IBM I worked as a lead application developer for a system based on J2EE technology. The system was a tool directed for the professional development of IBM employees, from a human resources perspective. I was basically responsible for both the lifecycle of the application and developing new features for it. While working in this project I got to work with major IBM products, such as DB2 and Rational Software Architect.

I designed and implemented applications for government agencies using Java EE platform. I was responsible for all phases of the development process, since requirement analysis, to database design and then implementation and testing.

Masters' research is focused on Entity Extraction techniques for portuguese language, based on deep neural networks and conditional random fields.