The voting process is the only thing that makes public trust in democratic systems possible. Traditional electronic voting techniques such as Electronic Voting Machines (EVMs) have limitations that could affect the results of elections. This paper introduces a transparent and secure electronic voting system, using Siamese Neural Networks (SNNs) for facial recognition and homomorphic encryption to protect data related to the voter. The implementation of the SNN model will ensure that voters are authenticated using face traits so that only eligible people may vote. Additionally, using homomorphic encryption, votes can now be encrypted without slowing down the system speed, and voter privacy is ensured throughout the voting process. This proposed method is more open and intended to provide a scalable option for major national elections.