I work as a data scientist in the Data Science and Advanced Analytics team.Currently focused on enhancing user experience personalization on the company's digital platforms.

I am a senior data scientist working in the company's analytics sector.My main objective is to contribute to the achievement of the company's financial goals through data consumption focused on approaches involving AI.Currently focused on the CRM area, I work on projects related to Shell Box.Shell Box is an application used at gas stations to register fuel supplies. In this app, the user can receive discount vouchers and accumulate points for performed transactions.In Brazil 2022, around BRL 6 billion were transacted in more than 20 million transactions inside Shell Box.Projects involved:• Promotional voucher distribution policy• Anti fraud• Customer segmentation• Churn forecast• Business indicators prognosticsResponsibilities:• Development of ML algorithms: the core of project solutions;• Create end-to-end pipelines for data ingestion, retraining, prediction and measuring of models;• Develop and calibrate drift monitoring services for model inputs in the retraining stage together with MLOps team;• Develop and monitor data validation services that train and consume the models in production stage together with the MLOps team;• Ensure maintainability, scalability and good CI/CD practices in developed pipelines;• Ensure healthy flows and good tracking practices for processes involving model training and consumption: versioning, releases, status changes (archieved, staging, production), logs, tags, etc.;• Ensure proper structuring of model outputs and generate visualizations/reports of project performance in according to customer demands.

Main goal: use data science to generate financial return for the company.Focus: recommendation engines via ML algorithms: collaborative and content-based filteringResponsibilities:- Modeling, testing, versioning and periodic updating of models;- Automation of ETL and ML processes to ensure maintenance and process execution time;- Monitoring results and generating reports;- Productization / operationalization of models when served in packages;- Work with MLOps team in model deployments when specific demands are required by the customer;- Assisting teams in the commercial sector to make data-driven decisions;- Case studies, feasibility analysis and ROI estimates for possible new opportunities with business teams.

Work carried out on historical data of industrial processes in the food sector. The aim was to optimize dairy production based on a concept: to produce just enough with the best quality possible and at the lowest cost.Activities:- Analysis/interpretation/reporting of data and business to understand how production relates to internal/external variables in all parts of the process;- Development of process-level solutions with their proper simulations and ROEs involved;- Provide assistance to the production planning and control sector via predictive models.Classifier, regression models and recurrent neural networks were used. Codes developed in python.

uni-paderborn.deThis project aimed to explore AI-based solutions that are potentially more efficient in the predictive maintenance program of companies whose mechatronics sector at the Universität Paderborn provides consultancy.The technical objective, in essence, consists of predicting, as accurately as possible, the remaining useful life of mechanical components submitted to constant stress, and, in case of failures, determining the dimensions and characteristics of the failures (cracks).The data used refer to readings from sensors installed in turbines. These were provided by NASA Ames' Prognostics Center of Excellence and are part of a competition called the PHM08 Data Challenge.Due to the noisy nature of the data, much of the time was devoted to a painstaking pre-processing work. The tsfresh python package was useful in the feature engineering step.Regression algorithms combined with time series were the basis of the solution.Work developed in Python, with support from R language and Matlab/Simulink.This work yielded my bachelor's thesis which title is: "Performance Study of Machine Learning Algorithms Applied to Prognostics and Condition Monitoring".Link for reading (portuguese): https://www.monografia.ufop.br/bitstream/35400000/2506/1/MONOGRAFIA_EstudoDesemecimentoAlgoritmos.pdfRepository (english):https://github.com/vrr187/Remaining_Useful_Life_Prediction

The work was carried out with the objective of developing solutions based on machine and deep learning applied to data from energy distribution networks.The challenges were divided into two:1. Detect fraud in energy consumption via unsupervised learning and anomaly detection techniques.2. Predict the next demands for energy supply as efficiently as possible. In this case, recurrent neural networks have been widely used, especially LSTM algorithm.Work done in Python.

www.gorceixonline.com.brThe work was developed in the Department of Analysis and Innovation (DEPAI) of the company in order to adapt the machinery of the mining sector to the current safety standards in the country: NR10 and NR12.Performed activities:- Study and feasibility analysis of possible solutions for equipment suitability;- Dimensioning and modulating the power circuits of the machines;- Projection of control panels using the AutoCAD Electrical.