Recommender systems is an information filtering mechanism designed to suggest products, services, and contents to users. This is achieved by analyzing past data, including user reviews and feedback, to offer personalized recommendations. Currently, recommendation systems are widely used on social platforms such as providing song and video suggestions; support online sellers or medical systems. Matrix factorization is the basic idea to predict a personalized ranking over a set of items for an individual user, is a widely used collaborative filtering method in recommender systems. This paper introduces a matrix factorization model with a neural network architecture. Initially, we establish a user-movie matrix that incorporates explicit ratings. Additionally, we use an embedding layer to generate implicit vectors which consists of embedding for both users and movies. Subsequently, we introduce a deep learning model aimed at identifying associations between users and products within the latent space, which represents users or products. Finally, we employ a loss function derived from the Mean Squared Error (MSE) to optimize our model and the goal is to minimize the above function value, thereby enhancing prediction accuracy. The experimental results achieve a 0.81 RMSE on the Netflix Prize dataset and 0.76 RMSE on the MovieLens dataset.