Designing a secure, scalable, and cost-effective framework for biometric data protection in virtual sizing systems : Case study of SizeWall

Virtual sizing platforms offer convenience in online retail, but they rely on sensitive biometric data such as facial scans and body meas-urements data that, once exposed, cannot be revoked. This thesis pro-poses a layered cybersecurity framework that protects biometric data across its entire lifecycle, from acquisition to storae, access, and machine learning inference. Using SizeWall as a case study, the frame-work addresses five critical areas: secure data capture, encryption at rest and in transit, multi-factor authentication, model protection, and real-time monitoring. Each layer is designed to minimize attack surfaces while preserving user privacy and system usability. The solution incorporates industry-standard cryptography (e.g., AES-256, TLS 1.3), privacy-preserving techniques (differential privacy, anti-spoofing), and continuous behavioral analytics to detect threats such as spoofing, model inversion, and unauthorized access. It is cost-effective, scalable to users, and compliant with regulatory standards. Evaluation criteria include privacy assurance, system speed, and adaptability to evolving threats. This research demonstrates how a defense-in-depth approach can be practically implemented in consumer-facing applications, offering a blueprint for secure biometric systems that are privacy-aware, performance-ready, and future-proof. The framework’s architecture is supported by academic literature and in-formed by real-world deployment constraints, making it relevant to both researchers and practitioners aiming to secure biometric technol-ogies in high-risk, data-driven environments.