Electronic voting technologies have emerged as a promising solution for improving transparency, efficiency, and security in modern electoral systems. However, the deployment of electronic voting platforms introduces significant challenges related to voter authentication, privacy protection, and system resilience in environments with unreliable infrastructure. This paper presents the Secure Hybrid Biometric Electronic Voting System (SHB-EVS), a novel architecture that integrates multimodal biometric authentication, edge computing infrastructure, and a zero-trust security framework. The proposed system combines fingerprint minutiae, facial recognition embeddings, and 3D liveness detection to ensure reliable voter identification while maintaining ballot secrecy through a reverse-flow voting protocol. The architecture employs hybrid data persistence using local edge databases and centralized cloud synchronization to ensure continuous operation during network disruptions. Experimental evaluation demonstrates that the multimodal authentication model achieves a combined false acceptance probability of approximately \(\:1.8\times\:{10}^{-9}\). Cryptographic micro-benchmarking shows that application-level encryption of a standard vote payload requires an average of 3.14 ms, while synchronization of 1,000 votes during network recovery requires approximately 11.45 seconds. Analytical Human-Computer Interaction (HCI) modeling indicates authentication completion times between 32.5 and 45.6 seconds across age groups, raising the projected System Usability Scale (SUS) score for elderly voters to an excellent 82. These results demonstrate that the SHB-EVS architecture provides a secure, resilient, and highly usable framework suitable for modern democratic electoral systems.