Compilers Researcher @ LAC


I am a MSc student in Computer Science at UFMG. I work with Fernando Magno in the Compilers Laboratory. I am being supported by a scholarship founded by Cyral.

My main interests are: Compilers, Programming Languages, Probabilistic Programming and Program Synthesis.


Master (MSc) of Computer Science (2019 - Now)

Institution Universidade Federal de Minas Gerias (UFMG), Brazil
Advisor Fernando Magno Quintão Pereira

Bachelor (BSc) of Computer Science (2014 - 2018)

Institution Pontifícia Universidade Católica de Minas Gerais (PUC Minas), Brazil
GPA 3.2/4.0
Grade First Class Honors ( Second place on PUC Minas Computer Science Undergraduate Thesis Competition )
Thesis NeurOMP: Automatic code paralellization using Reinforcement Learning
Advisor Luís Fabrício Wanderley Góes

Purposed a new Compiler Optimization, called NeurOMP, that uses the QLearning algorithm to automatic paralellize C/C++ code using OpenMP 4.0. Results shown that this method achieved the same speedup as a human specialist.


Honingstone: Building creative combos with honing theory for a digital card game

Journal IEEE Transactions on Computational Intelligence and AI in Games
Publication Date 2017/6

Presents a computational system, called HoningStone, that automatically generates creative card combos based on the honing theory of creativity. The system can generate combos that are more creative than a greedy randomized algorithm driven by a creativity metric.

A low-cost energy-efficient Raspberry Pi cluster for data mining algorithms

Conference European Conference on Parallel Processing
Publication Date 2016/8

Study of an unconventional low-cost energy-efficient HPC cluster composed of Raspberry Pi nodes when executing data mining algorithms compared with a Intel Xeon Phi. Results have shown that the Raspberry Pi cluster can consume up to 88.35% and 85.17% less power than Intel Xeon Phi when running Apriori and K-Means, respectively.

Algoritmo de Regras de Associação Paralelo para Arquiteturas Multicore e Manycore

Conference XVI Simpósio em Sistemas Computacionais de Alto Desempenho
Publication Date 2015/11

Development and evaluation of parallel version of the Apriori algorithm in OpenMP exploring the computational power of multicore and manycore architectures. The parallel Apriori achieves a performance gain of up to 10x on a multicore also on a manycore.


Neural Stemming

Language Python
Topics NLP, Deep Learning, Stemming

Usage of a Seq2Seq Neural network model to create a multilanguage stemming strategy.


Language Python
Topics Machine Learning, Automatic Parallelism, Compiler's Optimization

Group of Machine Learning based Compiler's Optimization.


Language C#, Ruby, Java
Topics Software Engineering, Startups

A Social Startup that aims to join people who want to do good, asociations who needed help and shops which wanted to colaborate to this ecosystem.


Language C++
Topics Graph Theory, Software Engineering

A general purpose library to work with graph problems.


Work Experiences


Job Title Data Scientist
Period 08/2018 - 03/2019
  • Optimized an industry manufacturing process, reducing their material cost, saving 8 million dollars. It uses XGBoost to predict relevant variables to the cost function, which is minimized using a Genetic Algorithm heuristic.

Job Title Data Science Intern
Period 09/2017 - 07/2018
  • Built a study content recommendation system based on students exams results. It uses several techniques and models used in Natural Language processing such as: TFIDF, Support Vector Machines (SVM) and Entity Tagging to try to improve the search engine, built using Elastic Search.

  • Developed a dynamic pricing application using Apache Spark to raise profits by 40% on retail segment. The developed systems automatically prepare the companies data, train and evaluates a Random Forest Models (to predict sales) using back test, most of the data are time series and finally it finds the best price to each product.

  • Optimized an e-commerce search engine using Apache Spark and a Vectorial Space Model.


Job Title Machine Learning Mentor and Reviewer
Period 02/2018 - 11/2018
  • Helped to teach more than 100 students how Machine Learning works and how to use it to solve real life problems. I have teach students about: Deep Learning (NN, CNN, RNN and GANs), Random Forests, SVMs, Linear Regression, Classification Trees (CART algorithm) and Reinforcement Learning.


Job Title Full Stack Web Developer Intern
Period 08/2016 - 09/2017
  • Projected and implemented a real-time chat app using Node.js, React and to help customers solve their doubts, problems and needs.

  • Developed a scalable backend infrastructure using Node.js, express, docker and MongoDB capable to handle a hundred orders a second.

Kukac Plansis

Job Title Artificial Inteligence Engineering Intern
Period 04/2016 - 08/2016
  • Built Speck, a Ruby on Rails solution based on IBM Watson Personality Insights and Tone Analyser APIs to personalize teaching experience based on student personality.

  • Made a application to help malls decide which shops to open. It was built using Node.Js and IBM Watson Tradeoff Analytics API. It also scraps for each brands Tweets and uses IBM Watson Tone Analyser to detected people emotion about it.

CreaPar Lab

Job Title Undergraduate Student Researcher
Period 02/2015 - 07/2018
  • CNPq Researcher Scholarship

  • Active participation on three research projects.

  • Three articles published.

  • One award won.