Flagship course
Fintech Fraud Model Audit Lab
A six-week cohort that takes one production-style fraud scoring context from sampling plan to a review-ready evidence pack.
Learning outcomes
- Design stratified samples that respect rare fraud events and operational volume
- Run challenge tests that isolate brittle thresholds and feature interactions
- Write residual-risk and limitation statements reviewers can trust
- Assemble an evidence pack with a navigable index and version notes
- Facilitate a cross-role readout that ends with clear follow-ups
Informational pricing
Signal Seat from ₩890,000 per learner. Bench Cohort packs available for teams. No checkout on this site — see Pricing.
Modules
1. Audit framing & scope locks
Define what the review covers, what it excludes, and how decisions get recorded.
2. Sampling that survives questions
Stratification patterns, rationale memos, and peer critique of your first draft pull.
3. Override rates & human loops
Connect model scores to analyst behavior without treating overrides as noise.
4. Challenge sets & threshold stress
Build adversarial cases and document what broke — and what did not.
5. Evidence pack assembly
Index, appendices, versioning, and a dry-run readout with facilitators.
6. Residual risk & follow-up cadence
Close the loop with owners, trackers, and a realistic refresh schedule.
Instructor
Sora Kim
Sora has led model risk reviews for payment and lending products across Korean fintech groups. She focuses on making audit artifacts readable for non-modeling stakeholders without stripping technical honesty.
FAQ
Common questions
Do I need production data access?
No. We use sanitized synthetic scenarios. You may optionally map exercises to your internal context offline, but never upload customer data to course tools.
Is this a coding bootcamp?
No. Light notebook literacy helps, but the emphasis is audit design, documentation, and cross-role communication.
What is a real limitation of this course?
We cannot certify your models for regulatory approval. Completing the lab improves your team’s review craft; it does not replace formal external audits or legal advice under Korean law.
How much time should I reserve weekly?
Plan for roughly five focused hours: one live session, one peer critique, and artifact writing.
Reviews from this course
“Module 4’s challenge log format is now our default. We still struggle with merchant-category edge cases — the lab acknowledges that gap instead of pretending a template solves it.”Yuri N. · Fraud analytics, Seoul
★★★★☆
Short take: “Sampling memo critique alone paid for the seat. Live sessions run brisk; catch-up recordings matter if you travel.”Verified learner