Built an Email/SMS Spam Detection system using NLP pipeline (tokenization, stopword removal, stemming, TF-IDF vectorization) Multinomial Naive Bayes achieved 97% accuracy and 94% precision on a 5,574-message dataset, outperforming SVM and Logistic Regression baselines.