%20(3).gif)
Timeline
Apr–Aug 2022 (5 months)
Team
1 PM (me), 1 designer, 3 engineers
My Role
End-to-end Product Management from product discovery, stakeholder alignment, privacy framework, and cross-functional execution
Identified 7.5 daily hours lost to payment friction, reframing as an opportunity to deploy responsible biometric+payment integration
- Rejected credit card infrastructure integration (2-3 years)
- Architected wage deduction integration (3 months) by separating databases, requiring verification, establishing opt-in consent
- Proving systems could integrate without breaking trust.
Eliminated ~15mins of payment friction time daily
Reduced dev timeline from 2~3 years to 3 months
Implemented Trust-by-Design for trustworthy systems
+ Featured in Korean Science documentary!! (Nov 2022)
Employees frequently forgot their credit cards when visiting the in-office robotic cafeteria, causing repeated interruptions and productivity loss during peak hours. While employees were accustomed to frictionless authentication (mobile pay, face unlock), the cafeteria, however, depended on physical cards, creating repeated friction at checkout.
~10 incidents during peak times (3-4pm), ~30 incidents per day
~15 minutes lost per incident
~7.5 hours of productivity lost daily
Always available! No card or phone needed
Already familiar enough through everyday use from mobile devices, airport check-in, banking kiosks
Appropriate for frequent, low-value internal payments
The real challenge wasn’t whether biometrics could reduce friction, but how to deploy them without compromising trust. Integrating biometrics with card networks would take 2–3 years and required external policy changes, unviable for an internal pilot. At the same time, biometric payments demanded zero tolerance for trust failures.
Integrating biometrics with card networks would take 2–3 years and require external policy changes, unviable for an in-house cafeteria.
A single false charge could permanently damage confidence in biometric systems.
Biometric data required strict compliance with Korean PIPA ('Personal Information Protection Act’), explicit consent, and strong governance.
Delivered the same user value (cardless payment)
Reduced delivery time from years to 1 month
Avoided external dependencies and policy risk
Instead of using existing card systems, I proposed deducting coffee payments directly from wages when verified through the biometric data. This could be only deployed in in-house systems, but after strategic considerations, there seemed to be more potentials and benefits, while still solving the core user problems. This decision reframed a technical limitation into a responsible deployment strategy.
Connected biometric authentication system to payroll database (only transmitting employee ID)
Biometric data are isolated from payroll systems and stored on premise as encrypted datasets.
Verification screen before every charge to avoid false payments
Explicit consent for authentication purposes during enrollment and 30-day deletion for inactive users
Defined industry-compliant biometric authentication threshold level of False Acceptance Rate for payment security
False Acceptance Rate: a security metric in authentication systems that measures the probability of an unauthorized user being mistakenly accepted as a legitimate user. Low is considered to be the secure system.
.png)
To User
To Business
[Special Feature] "Digital D.N.A." Part 2. Data.Network.AI. that makes money / YTN Science, YouTube Channel
Starting from: 18:27-Published: Nov, 2022Media: YTN Science, Korea
Instead of waiting years for card network integration, pivoting to wage deduction delivered user value immediately. This taught me that the "ideal" solution isn't always the right one, speed to value often beats.