Guilherme Resende Email & Phone Number

State Of Minas Gerais, Brazil

São Paulo, Brazil

- Studying prospects' data to unveil aspects that lead to accepting or denying them into our customer portfolio;- Developing, testing, and maintaining machine-learning pipelines to assess drawee credit risk, for both new and recurrent customers;- Recreating the company's main machine-learning pipeline to deal with increasing demands that are leading us toward a Series A round.

Belo Horizonte E Região, Brasil

- Worked on a project focused on reducing debt collection costs by using resources on customers with the biggest payment probability, given charging procedures, for one of the top 15 banks in Brazil. The project was implemented using Python and SQL. We used Kedro to structure and organize the pipeline, and Dask to process the data sources in parallel. This project improved the charging ROI by nearly 10%.- I also worked on a project to optimize resource allocation for Brazil's biggest port logistics operator. I had a late allocation in this project and my main contribution was analyzing the problem and noticing that a naive approach was applicable where only machine learning-based approaches were being considered.

Belo Horizonte E Região, Brasil

- Worked on developing a conversational assistant project for the biggest private bank in Brazil. We developed a Convolutional Neural Network solution with PyTorch and used it to classify conversation inputs and solve/guide user demands. My contributions ranged from model experimentation to performance analysis, and maintenance. - Also worked on an anomaly detection project focused on reducing costs regarding fraudulent requests for one of the biggest health insurance companies in Brazil. Here I experimented with unsupervised solutions (e.g. autoencoders and dimensionality reduction methods), Bagging approaches for supervised models, as well as maintaining the deep learning-based official pipeline.

Belo Horizonte E Região, Brasil

- Worked on assessing default patterns targeting customers with the most suitable charging procedures to avoid unnecessary costs. Here I performed data preprocessing, and analysis, and developed unsupervised approaches using PySpark inside an on-premise cluster.

Belo Horizonte E Região, Brasil

- Worked with bioinspired algorithms. In this project, I was responsible for experimenting with similarity functions to measure diversity rates in the population of an evolutionary algorithm. - Contributed to the development of an evolutionary approach to explore recommender functions.

Belo Horizonte E Região, Brasil

- Developing activities such as budget surveys for endomarketing products, communication, and benchmarking with other junior enterprises. Also worked leading a team towards the development of a site using WordPress with the Scrum methodology.

Master'S Degree, Computer Science

Thesis: On the bias of toxicity and sentiment analysis methods on the African-American English

Information Systems

Computer Technician