Run an Adversarial-Robustness Audit on a Face-Liveness Model for a Fintech
Overview
What this challenge is about.
You receive a stand-in face-liveness model with the same backbone as the production model plus a labeled evaluation set of 2,000 frames. Apply three standard digital attacks (FGSM, PGD, adversarial patch) at documented perturbation budgets, plus one printed-patch physical test (use a held-up phone screen as the attack surface). Report attack-success rate per attack, and propose two mitigations (input randomization, detection of patch-like inputs) with estimated impact. Write a 4-page risk-committee memo.
The Brief
What you'll do, and what you'll demonstrate.
Run a structured adversarial-robustness audit of the face-liveness model and propose ranked mitigations for the risk committee.
Earning criteria — what you'll demonstrate
- Implement standard adversarial attacks against vision models
- Design controlled audits that produce comparable results across attacks
- Reason about the gap between digital and physical adversarial attacks
- Communicate model risk to a risk-committee audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Career paths this builds toward
Canonical rolesAI Safety Researcher
Structured adversarial-robustness audits with risk-committee output are the canonical AI safety researcher deliverable in fintech and identity-product companies.
This challenge sharpens
- adversarial-robustness
- risk-reporting
- mitigation-design
Computer Vision Engineer
Understanding adversarial threat models is increasingly required for CV engineers shipping face or biometric models.
This challenge sharpens
- adversarial-robustness
- face-liveness
- pytorch
Research Scientist
Producing reproducible attack implementations and budget-controlled curves is the kind of research-rigor expected of a junior research scientist.
This challenge sharpens
- adversarial-robustness
- attack-implementation
- pytorch