In today's internet world the use of Machine Learning as A Service (MLaaS) is increasing day by day especially automated Machine Learning. In which users upload their dataset and auto Machine Learning (ML) will develop models after studying dataset. Due to this, the risk of compromising of privacy of users is also high. So, to secure data on a cloud a new approach is present in this paper. In which, we have been using homomorphic encryption (HE) based Rivest Shamir Adleman (RSA) algorithm to encrypt the data and Azure automated ML is used to develop models. Three different datasets are first encrypted by using the Homomorphic Encryption (HE) based Rivest Shamir Adleman (RSA) algorithm and then uploaded to the Azure automated Machine Learning (ML) platform. The accuracy of machine learning models developed using encrypted data is compared with machine learning models developed using unencrypted datasets.