A pressing issue for the modern era is dealing effectively with growing online hate speech cases, especially ones requiring scrutiny. This project presents a topic modeling integrated sentiment analysis model to analyze and categorize Twitter data. The data contains an English dataset in which the tweets were labeled hateful or non-hateful. Sentiment analysis was evaluated using four categorization algorithms. - Logistic Regression, Random Forest, Naive Bayes, and Support Vector Machine. These same algorithms are used to compare the proposed topic modeling feature extraction performance metrics. We also use a Tamil caste or immigrant hate speech detection dataset to evaluate the proposed method in a language other than English. Topic modeling uses a combination of three algorithms namely; Top2Vec, NMF, and BERTopic to uncover previously unknown topics within the hate speech corpus(cross-lingual), extract features from the text, and use machine learning algorithms for the predictive task.