- Monitoring the performance of credit models;- Proposing improvements in report analyses;- Assisting in the development of new models;- Interacting with other areas of Koin to aid in statistical studies;- Studying new partner's indicators to enhance the models.

Supervisor: Dr. Vinícius Ruela Pereira BorgesActivities: Studies in Natural Language Processing (NLP) for document similarity and document classification, employing both supervised and unsupervised approaches. Construction of crawlers for automatic data extraction and corpus building for annotation. Additionally, involvement in article writing and automation of administrative processes at the University of Brasília using machine learning and deep learning.

I analyze network indicators to identify opportunities for improvement.I use Descriptive Statistics to gain a better understanding of network data and generate insights.I develop automations using tools such as Python, Bash, Excel, Docker, and Power BI.I build ETL pipelines in Python for the construction of Data Warehouses.I have implemented Web Tools using Flask, Django, and Streamlit for network data analysis.I have created Power BI Dashboards for data visualization and Exploratory Data Analysis.I have modeled Relational Databases to support pipeline data in dimensional format.Results:- Over 4800% reduction in data load time for one of the company's projects.- Centralization of the company's data registry, increasing efficiency in data retrieval and reducing data redundancy.- Reduction of human effort in various company activities.

Activities: In this research project, I rotated through infrastructure/data science/development teams. Our client was the Court of Audit of the Federal District.I was the second author of a paper published at an international A4 conference. The main theme of this paper was the performance comparison between a graph-oriented database (Neo4j) and a relational database (MySQL) in contexts where the studied objects have many relationships with multiple entities.I was responsible for annotating the Text Corpus for the database used by the Natural Language Processing (NLP) models in the project.I conducted studies and summaries on Named Entity Recognition (NER) articles and Text Corpus construction related to the legal field.I created data visualizations in R and Python to compare the performance between relational and graph databases.I developed REST API endpoints with Flask to integrate the project's main platform with the databases.I managed Docker containers for application deployment.I used Jenkins and Github for Continuous Integration (CI/CD).I modeled data with Neo4j for storing data that fed the Named Entity Recognition (NER) models.I managed and configured Linux machines and servers.Results: In addition to an accepted paper at the CISTI international conference indicating the significantly superior performance of a graph database over a relational one in the mentioned context, we also contributed to the transparency of government tenders in the Federal District with visual tools and the availability of datasets from our annotations.These datasets can be used by technical professionals, for example, to implement anomaly detection models for Bid Fraud.

Activities: I developed Python scripts for task automation.I was one of the responsible parties for annotating Text Corpora for NER models.I conducted studies on Artificial Neural Networks with PyTorch.I conducted experiments and evaluations of NER models for classification and segmentation of Official Documents from the Official Gazette of the Union (DOU) in Brazil.I studied the construction of relationships and nodes in a graph-oriented database (Neo4j).

Activities: I was one of the main individuals responsible for managing access for large companies within the CNP Asurrances group in the Brazilian insurance scene, including CNP Brasil (Caixa Seguradora), Youse Seguros, Previsul, and Odonto Empresas.Managed the email inbox of the company's security department.Conducted Phishing analyses based on email headers to detect malicious emails.Automated repetitive tasks using Python.Activities: I was one of the main individuals responsible for managing access for large companies within the CNP Asurrances group in the Brazilian insurance scene, including CNP Brasil (Caixa Seguradora), Youse Seguros, Previsul, and Odonto Empresas.Managed the email inbox of the company's security department.Conducted Phishing analyses based on email headers to detect malicious emails.Automated repetitive tasks using Python.

Special tutor at the PTE in the Network Fundamentals subject, preparing summaries, answering questions and giving weekly classes to the tutee student on the content related to the CompTIA Network+ certification and the subject