Advising a lender off a DIY capture flow and onto a compliant validation path
The problem
Fast Credit built its own selfie capture UI to avoid SDK lock-in, submitting images via raw REST calls. That meant no device metadata, rooted-device checks, or encrypted payload — fraud liability sat entirely with them. Separately, their Tier 2 upgrade flow reused the same onboarding selfie for a second verification step, and their ID-card flow relied on a document type that had been discontinued.
My approach
Rather than pushing a full SDK rewrite the team wasn’t ready to commit to, proposed the smallest safe change: adopt just the capture component with a skip-submission flag, preserving their existing backend architecture while restoring the fraud-detection layer. For the discontinued document type, mapped a path through database-level biometric validation instead of the deprecated document-scan route.
What I built
A phased integration recommendation (capture-component adoption, fresh-selfie-per-step policy, database-level ID validation) delivered as a joint technical review with the partner’s engineering and product leads, replacing an ad-hoc support relationship with a documented plan.