Work History
Sep 2010 - Present

Data Scientist (Market Intelligence Analyst)


(Multinational company focused in Agribusiness and Risk Management. The largest private company in America)

* Develop, implement and test the machine learning algorithm (IDL language) for the Crop Recognition and Automatic Determination of the Planting Date project which analyzes satellite images in order to determine planting dates and the crops planted in producing regions. * Design and conduct freight market data analysis. Create and maintain predictive models for the freight rate index and present it in weekly meetings to business unit leadership and 30 more people. The models helped Cargill manage logistic risks during a difficult period of changes in the Brazilian freight market.

* Performed complex data-driven analysis in problem solving (worked in a pair). The suggestions provided by the analysis were used and implemented at Cargill's terminal to increase its logistic productivity. Manage multiple projects simultaneously in two different areas: Logistics and Global Trading Analytics.

In these projects, I work with:

  • Fourier Transform and Wavelet Transform (IDL Language), used as filter techniques and as feature extraction.
  • K-Nearest Neighbor-KNN (IDL Language): Matching strategy.
  • Multivariate Linear Regression (R and Excel): Predictive model.
Jul 2010 - Aug 2010

Physicist - C/C++ programmer


(Startup focused on digital marketing and technology)

Specialist in pattern recognition and image/signal processing. I worked on these projects:

  • Wrinkle analysis of facial skin: In this project, I developed an image processing algorithm to extract facial wrinkles. A score was attributed to the wrinkles and these data were correlated with medical data in order to determine skin age. It was a marketing project that was being developed for a cosmetic company to recommend products to their clients based on their skin age.
  • The Footprint project team developed and built a footprint scanner. I worked as a contributor in this project by providing suggestions to reduce light noise effects. I also started the development of the pattern recognition algorithm to determine the arch type of the foot based on the images acquired by the scanner. In these projects, I used OpenCv library (C/C++) and these image techniques: Gabor was used as a kernel for a convolution filter.
Jan 2008 - Apr 2010

Physicist - C/C++ and Matlab programmer

FIT - Fine Image Technology

(Startup focused on technology and Magnetic Resonance Imaging)

I worked with statistics and signal processing on the projects below:

* Automatic speaker recognition project aims to recognize individual people (verification task) in a data base from their voices. I worked on the machine learning and also on the preprocessing voice separation algorithms.

* Managed and coordinated with a colleague the implantation of a speaker recognition algorithm in the client company. The algorithm developed in Matlab was ported to JAVA language.

* Magnetic noise attenuation system: In this project, I designed a compensation coil in order to reduce magnetic noise in magnetic resonance scanner room. Also, I worked on the implementation of the attenuation algorithm.

* Demonstrated leadership skills by managing two colleagues in the construction of a demonstration version of Radio frequency shielding exhibited at the International Congress of Radiology-2009. Managed a third-party team (5 persons), planned the project and its interfaces with other related teams on the same project.

In these projects, I used these mathematical and statistical techniques:

  • Fourier Transform was used to compute the Mel-frequency Cepstrum coefficients.
  • Gaussian Mixture Model, parametric approach, was used to fit the empirical distribution of voice attributes.


Jul 2012 - Jul 2012

Short program

Massachusetts Institute of Technology

Data and Models in Engineering, Science, and Business (Classroom)

2005 - 2008

Master's Degree

Universidade de São Paulo

Dissertation titled: Three-Dimensional Simulation of the Accommodation Process of the Human Eye. Development in C/C++.

Number of visits/downloads up to 2015/06: 702/361

Developed in C++ (OOP) of the Veye framework. Worked on the architecture,
on the verification routines and the validation of the project.

2000 - 2003

Bachelor's degree

Universidade de São Paulo - IFSC

Key disciplines: Algorithm and Data Structure I,
Object-Oriented-Programming,Statistics I, Image Processing and
Computational Vision, Parallel Programming and Methods in Computational


Dec 2013 - Dec 2013

Data Analysis - Johns Hopkins Bloomberg School

Jun 2013 - Jun 2013

Introduction to Data Science - University of Washington


Text Section