For sentiment analysis, the importance of user opinions in our digitally globalized area cannot be overemphasized. How optimal deep learning techniques could be further improved by using Convolutional Neural Networks (CNNs) in conjunction with Multi-layer Perceptrons (MLPs) for an effective sentiment classification system has been discussed in this paper. Bringing CNNs’ ability to detect hierarchical patterns in text and MLP’s capacity to model complex, non-linear relationships together, the study introduces a hybrid CNN-MLP model. This experimental use of the hybrid model with IMDb movie reviews has established its considerably better performance over single CNN, MLP, and the traditional Logistic Regression and Gaussian Naive Bayes methods, with an impressive accuracy of 96%. The advantage of the hybrid framework lies in the combination of features and classification, making it possible to employ the model more frequently in text-centric jobs Real-world applications, including real-time sentiment evaluation, content moderation, and personalized recommendations, are particularly fostered by the methods this study has tested. In fact, results show how much efficient and adaptive the model is for Sentiment Analysis in many other datasets.