Biometric Authentication Ecosystem

The Challenge

e& UAE operates 40 retail stores across the country — plus 2 fully autonomous stores with no human agents — each processing thousands of customer transactions daily. Every transaction requires identity verification per UAE regulatory requirements (ICA — Federal Authority for Identity, Citizenship, Customs & Port Security).

The existing verification process was manual, slow, and costly. The challenge was threefold: deploy face recognition across all retail touchpoints (stores, mobile app, SAS self-service machines, and autonomous stores), launch a voice biometric system branded "My Voice Is My Password," and integrate fingerprint verification — all while maintaining ICA compliance and data privacy standards.

My Role

As AI Product Owner, I was responsible for:

• Defining the biometric product roadmap across voice, facial recognition, and fingerprint modalities
• Owning the backlog and prioritizing features based on regulatory requirements, store feedback, and cost-reduction targets
• Coordinating between biometric technology vendors, internal IT infrastructure teams, and retail operations
• Ensuring compliance with ICA regulations and ISO/IEC 42001 ethical AI governance
• Designing the end-to-end customer journey for in-store biometric verification

Process

1. Research & Requirements

Conducted field research across retail stores to understand the current verification workflow, pain points, and peak-hour bottlenecks. Interviewed store managers, frontline staff, and compliance officers to map requirements.

Key findings: verification took an average of 3-5 minutes per transaction, document-based checks were error-prone during rush hours, and customers frequently complained about wait times during SIM activation.

2. The Three Products

Face Recognition (Primary product): The largest of the three. Deployed across 40 retail stores, a dedicated mobile app, SAS self-service machines, and — most notably — 2 fully autonomous stores operating with zero human agents. Customers walk in, verify identity via face recognition, complete their transaction (SIM swap, plan upgrade, device purchase), and leave. 3M+ users served.

Voice Biometric — "My Voice Is My Password": Branded voice authentication system where customers register their voiceprint and use it as a password for future interactions. Deployed across call center and IVR channels — customers authenticate by speaking naturally rather than answering security questions. Reduced call handling time and eliminated the frustration of forgotten PINs.

Fingerprint Verification: Traditional biometric for high-security transactions requiring physical presence confirmation. Integrated with ICA systems for regulatory-compliant identity verification at retail counters.

3. Pilot & Iteration

Launched a pilot across 5 high-traffic stores to validate accuracy, speed, and customer acceptance. Iterated on the UX based on real-world feedback — adjusted camera positioning for facial recognition, optimized lighting guidance for staff, and refined the voice enrollment flow to reduce drop-off.

4. Scale to 40 Stores + Autonomous

After pilot validation, rolled out face recognition across all 40 stores, SAS machines, mobile app, and the 2 autonomous stores. The autonomous stores were the most ambitious deployment — requiring the biometric system to handle the entire customer journey without any human fallback. Coordinated hardware deployment, staff training, and integration with CRM and ICA verification APIs.

5. Governance & Compliance

Implemented comprehensive AI governance aligned with ISO/IEC 42001:

• Data privacy: Biometric data encrypted at rest and in transit, with clear customer consent flows
• Bias testing: Regular audits to ensure facial recognition accuracy across diverse demographics
• Transparency: Customers informed about biometric data usage with opt-out alternatives always available
• Regulatory compliance: Full alignment with ICA identity verification standards

Key Takeaways

• Multi-modal is essential — no single biometric works in every environment. Graceful fallback between modalities is what makes the system reliable in the real world.
• Pilot small, scale fast — 5 stores gave us enough signal to iterate before committing to 42. The pilot caught issues (lighting, noise) that lab testing missed.
• Compliance is a feature — in regulated industries like telecom, building governance into the product from day one accelerates deployment rather than slowing it down.
• UX matters for AI too — my design background was critical in making biometric verification feel seamless rather than intrusive to customers.