Works
Case studies in trust.
Three projects in trustworthy systems: earning trust from people, making it portable across tools and compliances. From biometric payments to airport identity to responsible AI inside research admin. The shipped product is the visible half; the judgment behind it is the rest.
Trustworthy AI for Research Budgets
Responsible AI for research-award budgeting, embedded inside UW's SAGE.
Narrowed an "automate award setup" mandate into one diagnosis (grant managers build every budget twice) and translated HAX, PAIR, and AI4RA into trust-calibrated AI inside SAGE. Projected 75–150 hrs per grant manager per year (3,000+ org-wide).
Pay with Your Biometrics
Cardless biometric payments, shipped in 1 month with zero false charges.
Proving trusted systems could connect without eroding confidence in either, recovering 1,270+ hours a year with 0 false charges (KFTC).
One Palm, Banks to Airports
Korea's first cross-industry biometric framework: one palm enrollment, 12 banks, 14 airports.
Cross-industry biometric interoperability across 12 banks and 14 airports. 500K+ travelers onboarded in year 1, 0 security incidents (KFTC).
Side builds
shipped solo, on my own time
Keep exploring
You just spent time in Works
Observatory
See what I'm building →
Monthly digests, side projects, and what didn't make the case studies.
Writings
Read the essays →
Notes on human-AI interaction, security UX, and the discipline of working in public.
Building something where trust is the hard part? Write me about the constraint that is costing you. jiyaechoi19@gmail.com