Master of Computer Science (Ago 2011 - Feb 2014)
- Title: Low False Positive Learning with Support Vector Machines.
- Advisor: Prof. Dr. Anderson Rocha.
- Member of Recod Reasoning for Complex Data.
Bachelor of Computer Science (Ago 2007 - Jul 2011)
- Title: Automatic Relevance Ranking of Offers Based on their Features.
- Advisor: M.Sc Fernando Henrique Bezerra Cardoso.
- Highest MGP (GPA equivalent) in class.
- Algorithms and Data Structures I TA.
Selected Work Experience
Data Scientist and Lead Software Engineer (Mar 2014 - Feb 2016)
- Developed a method that detects porn audio with an accuracy of over 89.5%.
- Optimized and ported our learning pipeline to a mobile device, with an speed gain of 63.5%.
- Developed and delivered real time porn and violence detection tools for Android and Linux.
- Our team got the second place at the VSD generalization task of MediaEval 2014.
- The research resulted in a patent.
Technologies: C++, Python, Android, Shell Script, OpenCV, LibSVM, FFmpeg, openSMILE, VLFeat.
Software Engineer (Dec 2010 - Jul 2011)
- Developed a distributed module that synchronizes every day hundreds of files in real-time between two Plone-based websites.
- Developed a module to automatically import thousands of documents to a Plone-based website.
Technologies: Python, Plone.
Software Engineer, Intern (Mar 2009 - Apr 2010)
- Developed several features to an MMORPG browser game, including pathfinding, groups and email management.
- Developed The Giant Robot, on a team of 1 enginner and 2 designers in 3 months.
- Developed Axe Penny Paintball, on a team of 3 enginners and 1 designer in 1 month.
Software Engineer, Intern (Feb 2008 - Feb 2009)
- Implemented and launched several features to an student information system.
- Developed and delivered a system to register employee performance.
- Developed and launched several web applications and websites.
Research Project (Mar 2014 – Jun 2014)
- Developed a method that detects crowds of people on images, on a team of two researchers.
- Accuracy of over 94%, based on a dataset of 660 images collected from Google Image Search.
- Research conducted under supervision of Prof. Dr. Siome Goldenstein and Prof. Dr. Sandra Avila.
Technologies: Python, Shell Script, OpenCV, LibSVM, BinBoost.
Research Project (Dec 2012 – Feb 2014)
- Novel 2-level learning method for low false positive classification.
- Methods effectiveness showed trough comparisons to other solutions in 33 datasets.
- Research conducted under supervision of Prof. Dr. Anderson Rocha and Prof. Dr. Jacques Wainer.
Technologies: Python, LibSVM.
Research Project (Aug 2012 – Dec 2012)
- A product ranking algorithm that takes into account the cost-benefit ratio of the products.
- Agreement of 64% according to a test conducted with 79 users.
- Research conducted under supervision of Prof. Dr. Anderson Rocha.
Course Project (May 2012 – Aug 2012)
- A HTML spider that automatically collects products data from Google Product Search and Buscapé.
- Collected and refined data of 495,000 offers from 34,000 laptops from Google Product Search during 7 months.
Lead Programmer (May 2013 - Present)
- Short course presented at Tiradentes University.
- Experienced with GNU/Linux, Git, VIM, LIBSVM, and LaTeX.
- Good with Node.js, React/Redux, Java/Android, and Shell Script.
- Good with FFmpeg, R, OpenCV, Apache, PostgreSQL, and MySQL.
- Can effectively communicate in English.