Stanley Paul

Customer Reviews Sentiment Analysis

Designed and implemented a full NLP pipeline for Customer Reviews Sentiment Analysis, including text preprocessing (cleaning, stop-word removal, lemmatization), feature extraction (TF-IDF, BERT embeddings), and supervised model training (Logistic Regression, SVM, XGBoost, deep learning models). Conducted exploratory text analysis to identify common themes and sentiment drivers. Applied hyperparameter tuning and cross-validation to maximize performance, achieving ~80% accuracy and ~75% F1-score. Delivered insights through visualizations and dashboards, enabling stakeholders to monitor sentiment trends and prioritize product improvements.

Churn Customer Prediction

Designed and implemented an end-to-end Customer Churn Prediction pipeline model to help identify high-risk customers for proactive retention. Conducted data preprocessing (handling missing values, encoding categorical variables, scaling numerical data), exploratory data analysis (visualizing churn drivers such as tenure, contract type, charges), and engineered predictive features. Trained and compared ML models including Naive Bayes, K-Nearest Neighbors, Random Forest, and Gradient Boosting and Multi-Layer Perceptron. Performed hyperparameter tuning (GridSearchCV) and model calibration (Threshold) to optimize performance. Achieved ~40% improvement in recall for churned customers. Delivered insights to stakeholders through dashboards (Tableau/Matplotlib) and translated predictions into actionable business recommendations.

• ● Delivered multiple embedded and desktop software solutions for Safran Electronics & Defense. Optimized processing speed by 40% and reduced manual errors by 35% through automation and robust application design.

• ● Developed integration tests, embedded simulations, and parsing programs, fulfilling 100% of client requirements.

• ● Collaborated with cross-functional teams and managed software documentation, testing, and version control.