cv

Consultant / Research Assistant

The World Bank

I worked in developing, applying, fitting, improving, and optimizing econometric models for geospatial land use data in Latin America and the Caribbean (LAC). I handled an encrypted dataset comprising tens of millions of entries, for which I was also responsible for data cleaning and processing, focusing on scalability with PyTorch and Nvidia CUDA. This assignment resulted in a submission for the Nature Sustainability journal.

• PyTorch
• Statistics
• Econometrics

MS. in Applied Mathematics and Data Science

School of Applied Mathematics, Getulio Vargas Foundation, Rio de Janeiro, Brazil

Applied Mathematics

• Real Analysis
• Numerical Analysis
• Computational Statistics
• Causal Inference
• Stochastic Calculus
• Statistical Inference
• Quantitative Biology

BS. in Data Science and Artificial Intelligence

School of Applied Mathematics, Getulio Vargas Foundation, Rio de Janeiro, Brazil

Data Science

• Real Analysis
• Numerical Analysis
• Computational Statistics
• Causal Inference
• Stochastic Calculus
• Statistical Inference
• Quantitative Biology

ICIAM Fellowship CNMAC 2024

International Council for Industrial and Applied Mathematics (ICIAM)

Work: 'Shape-contrained function emulation with Gaussian processes'

Scientific Initiation scholarship

National Institute of Mathematics Science and Technology (INCTMat)

Project: 'Shape-constrained Gaussian Processes'

Silver Medal

Elon Lages Competition, Brazilian Mathematical Olympiad (OBM)

Gold Medal (1st Place)

Science, Technology and Innovation Fair of the State of Rio de Janeiro (FECTI)

Silver Medal

Brazilian Mathematical Olympiad of Public Schools (OBMEP)

Bronze Medal

Brazilian Mathematical Olympiad of Public Schools (OBMEP)

Open Labeller for Iterative Machine learning

OLIM is a user-friendly labeling interface designed for individuals without prior data science knowledge. Its primary use case is to assist in labeling patient-related data extracted from medical texts. Future versions aim to support a wider range of data formats and domains.

• Active Learning
• Natural Language Processing
• Machine Learning

Shape-constrained Gaussian Processes

The use of Gaussian processes is a powerful technique for estimating functions which cannot be evaluated across their entire domain, allowing for the quantification of estimation uncertainties, especially when following the Bayesian paradigm. However, when we have prior knowledge about the shape properties of a function, especially its derivatives, it is possible to enhance the quality of our estimates. This project focuses on the development and application of shape-constrained Gaussian processes to improve function estimation. Supervised by Professor Luiz Max de Carvalho, PhD.

• Gaussian processes
• Bayesian statistics
• Shape-constrained estimation

A Bayesian approach to understanding the Homicide Rate in the City of Rio de Janeiro by administrative regions through their Social Progress Index indicators

This study aims to investigate the relationship between the homicide rate in the city of Rio de Janeiro and the indicators of the Social Progress Index. Our approach involves employing Bayesian methodology to estimate the parameters of three multilevel models and subsequently comparing their performance. The Social Progress Index serves as a measure of the overall quality of life and social well-being of the population, and it has been regularly published by the Pereira Passos Institute for the City of Rio de Janeiro biennially since 2016. Given the well-known issue of violence in Rio de Janeiro, the homicide rate serves as a pertinent indicator of this problem. The city is divided into 33 administrative regions, and we utilize the corresponding data throughout this research.

• Bayesian statistics
• Multilevel models
• Social Progress Index

Gaussian Processes Visual Tool

This paper presents the Gaussian Processes Visual Tool, an interactive tool for visualizing Gaussian Processes (GPs). The tool combines the power of Svelte, D3.js, Flask, and GPyTorch to provide a flexible and user-friendly interface. With the Gaussian Processes Visual Tool, users can visualize and understand GPs, make observations, and explore various kernel functions. The tool offers simulation options, allowing users to set axis parameters, likelihood variance, and even upload custom datasets. It stands out for its aesthetic design, comprehensive functionalities, and the integration of multiple important technologies. The Gaussian Processes Visual Tool fills a gap by offering a beautiful and useful tool that combines all essential GP functions in one place.

• Gaussian processes
• Bayesian statistics
• Interactive visualization

Voyager’s Journey: an interactive visualization of NASA’s mission through the solar system

Joint work with Juan Belieni (FGV EMAp) and Marcelo Amaral (FGV EMAp). TThis article presents an interactive visualization using web technologies of the Voyager program by showing, for each probe, a simulation of the Solar System, containing the probe’s position for each day, the main events of the mission inside a timeline and a gallery of images captured by them. The visualization is controllable by a player, which offers the option to play, pause, change the speed and reset the whole visualization.

• D3.js
• NASA's Voyager program
• Interactive visualization