Guilherme Tolentino

I build machine learning, NLP, and generative AI solutions while bringing 15+ years of analytical work in high-complexity environments. My work combines Python, SQL, statistical modeling, automation, and clear communication for technical and executive stakeholders.

In my sentiment-analysis project, I built an end-to-end NLP pipeline for movie reviews and compared TF-IDF with Logistic Regression, LightGBM, and BERT. The BERT-based model achieved an F1 score of 0.88, with a focus on reproducibility, data quality, interpretability, and GitHub documentation.

For a telecom churn-prediction project, I led exploratory analysis, feature engineering, and model comparison across Logistic Regression, Random Forest, LightGBM, and XGBoost. The best XGBoost model reached a ROC-AUC of 0.85 and supported retention-focused business decisions.

At Fundação de Desenvolvimento Gerencial, I create analytical reports and performance indicators for executive decision-making. Previously, I automated ERP data flows, reducing operational time by 80%, led 17 professionals, and managed performance analysis for a portfolio of 230+ clients.