Google Brain / Google Research

Research Intern, Remote (May 2021 - Aug 2021)

• • Machine Learning for Causality Research Project: We created a Deep Learning-based framework to generate new benchmark datasets for Causal Inference analysis. The proposed benchmark contributes to the challenge of evaluating treatment effect estimators and targets applications with unstructured confounding by using images as covariates.

  • I was responsible for: implementing a reproducible, robust, and easy-to-use version of the framework in Python (code), running the experiments on Google Cloud, writing the paper, and presenting the work at a workshop.

Borealis AI - RBC

Machine Learning Research Intern, Remote (Sep 2020 - Dec 2020)

• • Multi-task learning Research Project: We proposed modifications to an existing multi-task learning architecture to improve predictions. These modifications were evaluated in three datasets. It was shown to be especially useful in applications with very heterogeneous tasks.

  • I was responsible for: implementing a reproducible version of the proposed method in Python + Pytorch, the data pre-processing and implementation of the baselines (code), running the experiments in a remote cluster, writing several sections of the paper, and presenting the work at a workshop. (code, paper).

Simon Fraser University

Research Assistant, Burnaby, Canada (Sep 2017 - Present)

• • Pharmacogenomics and Precision Medicine: Partnership with the BC Children's Hospital. Responsible for investigating causal associations between SNPs and side effects of drugs used in therapies to treat cancer in infants.

  • Driver Gene Discovery: Responsible for data pre-processing, analysis, literature review, development of models (causal discovery models, factor models, probabilistic graphical models, predictive models), and writing.

Simon Fraser University

Teaching Assistant, Burnaby, Canada (Apr 2018 - Aug 2019)

• • Tutoring, lab assistance, and marking;

  • Courses: Introduction to Computer Science/Programming I, Computational Data Science, and Artificial Intelligence.

Hekima

Data Scientist, Belo Horizonte, Brazil (Feb 2017-Aug 2017)

• Data Science Projects: Development and implementation of data science projects for a diverse portfolio of clients.

• I was responsible for cleaning the data and using statistical and machine learning tools to provide relevant insights. I also create data visualizations, reports, and presentations for the clients.

• Startup Hekima was acquired by iFood in January 2020;

Prograd (Undergraduate Dean's Office) - Federal University of Minas Gerais (UFMG)

Statistician Intern and Statistician, Belo Horizonte, Brazil (Oct 2013-Feb 2017)

• The project focused on the development of strategies to decrease university dropouts and the impact of socio-economic policies adopted by the university;

• Automated Reports Project: The goal was to generate several PDF files with plots, metrics, and insights with students' metrics for all university departments. These reports were sent to the (approximately) 86 departments twice a year. I collaborated to create the automated reports, which were developed using R Markdown, R, and Latex. The automation reduced the work time to make these reports from months to a week.

• I was also responsible for cleaning the data, analyzing the data with the appropriate statistical methods, and creating data visualizations to be shared with the university's departments and the general public in several other smaller projects.

[1] [Paper] Aoki, R., Ester, M.. "Causal Inference from Small High-dimensional Datasets", under review, 2022. (link)

[2] [Paper] Aoki, R., Chen, Y., Ester, M.. "M3E2: Multi-gate Mixture-of-experts for Multi-treatment Effect Estimation", under review, 2022. (link)

[3] [Poster] Aoki, R., D’Amour, A., McLean, C., Yadlowsky, S.. "ICETEA Benchmark: Semi-synthetic Treatment Effect Methods Evaluation with Images as Covariates." American Causal Inference Conference (ACIC), 2022. (code)

[4] [Paper] Aoki, R., Tung, F., and Oliveira, G.. "Heterogeneous Multi-task Learning with Expert Diversity." IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2022. (paper, slides, code) Note: This work was previously presented at the 20th International Workshop on Data Mining in Bioinformatics (BIOKDD), 2021.

[5] [Paper] Aoki, R., and Ester, M.. "ParKCa: Causal Inference with Partially Known Causes." Pac Symp Biocomput. 2021. (link, Supplemental Material)

[6] [Poster] Aoki, R., and Ester, M.. "Bayesian Predictive Model combined with Matrix Factorization for Causal Inference Analysis." 14th Machine Learning in Computational Biology (MLCB) meeting, co-located with NeurIPS. 2019. (link)

[7] [Paper] Aoki, R., Assuncao, R., Melo, P.. "Luck is hard to beat: The difficulty of sports prediction." Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2017. (link, video) Blogs Comments: marcodena, acolyer, reddit.

[8] [Poster] Aoki, R., Assuncao, R., Melo, P.. "Measuring the Size of the Surprise Box in Soccer Leagues."(Original Title: “Medindo o tamanho da caixinha de surpresas em ligas de futebol.”), 31rd BrazilianSymposium on Databases (SBBD), 2016. (link)

[9] [Poster] Aoki, R., Soier, A., Fiqueiredo, D.. "Analysis of Political Knowledge of University Students Using Item Response Theory." (Original Title: “Análise do conhecimento político dos universitários utilizano Teoria de Resposta ao Item”, 21rd National Symposium on Probability and Statistics (SINAPE), 2014.

Doctor of Philosophy (Ph.D.) in Computer Science

Simon Fraser University, Burnaby, British Columbia, Canada (Sep 2017 - Present)

• • Supervised by Prof Martin Ester.

  • Project: Causal Discovery for Computational Biology.

  • Cumulative GPA: 3.89/4.33 (89.84%)

  • Research Areas: Machine Learning, Causal Inference, and Computational Biology.

Master of Science (MS) in Computer Science

Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil (Mar 2015 - Apr 2017)

• • Supervised by Prof Renato Assuncao and co-supervised by Prof Pedro O.S. Vaz de Melo.

  • Project: Analysis of luck's influence on sport

  • Cumulative GPA: 3.88/4.5 (86.22%)

  • Research Areas: Machine Learning, Bayesian Statistics and Sport Analytics.

  • Teaching Assistant: Probabilistic Graphical Models (2016)

Bachelor of Science (BS) in Statistics

Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil (Mar 2011- Dec 2014)

• • Supervised by Prof. Fabio Demarqui.

  • Cumulative GPA: 3.6/4.5 (80.00%)

  • Research Areas: Traditional Statistics, Bayesian Statistics and Regression Models.

  • Undergrad Research Assistant in the Statistics Department with Prof. Sokol Ndreca (Aug 2012 - Sep 2013).

  • Undergrad Research Assistant in the Mathematics Department (Mar 2011 - Jul 2012)

• SFU Graduate Fellowships - 2018, 2019, 2020, 2021.

• D.E. Clark Annual Graduate Award in Computing Science - 2019 (link): This award recognizes and provides funding to graduate students who have demonstrated positive contributions to the Women in Computing Science (WICS) student group, including leadership, volunteerism, and/or ambassadorship to the external community.

• Graduate-Undergrad Research Mentorship Award 2019 (link) to attend the International Conference on Machine Learning (ICML).

• Bernardo Alvares Scholarship - 2013: This award is given to the student with the highest GPA in the Statistics Department.

• Scholarship from National Council for Scientific and Technological Development (CNPQ) from March 2012 to July 2012 to develop stronger mathematical foundations during the undergrad (link).

• Bronze Medal in the Brazilian Mathematical Olympiad of Public Schools (OBMEP) in 2007 and 2010 (link).

Ester Lab - SFU

Lab Communication Manager and Lab Meetings Manager (2021, 2022)

• Maintain and update our lab website (link);

• Plan and organize weekly lab meetings;

• Recruit and communicate with speakers of our lab meeting presentations.

Women in Machine Learning (WiML) Workshop co-located with the virtual NeurIPS

Virtual Logistics Co-Chair, Remote (2020)

• Event with more than 1000 attendees from academia and industry;

• I recruited, communicated, trained, scheduled, and managed volunteers and super-volunteers to help on the workshop day.

Invent the Future - AI4All

Program Mentor and Project Mentor, Burnaby, Canada (Aug 2018, Aug 2019, July 2020, July 2021)

• Two-week summer camp focused on bringing Artificial Intelligence (AI) expertise, community, and mentorship to trans and cisgender women students in Grade 9 - 12.

• As Program Mentor, I collaborated on creating the schedule, setting up Google Classroom, and organizing the Machine Learning content (slides, videos, Google Colabs). I also created tutorials to enable students to properly set up their machines.

• As Project Mentor, I was responsible for the Computational Biology project. Along with my co-project mentor, we prepared a week of content on the project.

• In the presential years, I also supervised the students during programming exercises and off-campus trips.

Women in Computer Science (WiCS)

Grad Coordinator and Mentor, Simon Fraser University, Canada (Oct 2018 - Aug 2020)

• Organization of events and workshops to support and encourage women in Computer Science.

• Mentor of new graduate students in 2018 and 2019.

OMICS group

Organizer and Finances, Simon Fraser University, Canada (Oct 2018 - Aug 2021)

• Part of the Research Day organization team. The full-day event featured oral presentations, poster presentations, and a keynote speaker on data-driven biology and the interdisciplinary fields that stem from it. This event was open to SFU students, staff, faculty, and alumni.

• I contributed to organizing monthly meetings, workshops, finances of the Research Day, and other events.

Conferences Volunteer

• ICML - 2020

• NeurIPS - 2019

• Women in Machine Learning Workship (WiML) - 2019

• Grace Hopper Celebration (GHC) - 2018

EstatMG

Statistician and Marketing Director, Belo Horizonte, Brazil (May 2013-Oct 2014)

• Local non-profit organization entirely managed by students.

• Our mission was to provide data analysis for individuals and companies that could not afford to pay for regular services;

• As a Statistician, I was responsible for cleaning the data, creating descriptive analysis and visualizations, and using a wide range of statistical methods, such as regression models, sampling methods, quality control, clustering, market survey, and principal component analysis.

• As a Marketing Director, I worked to increase brand awareness, improve recruiting numbers, and expand the number of clients.