The situation.
When onboarding declarations, risk decisions and monitoring controls live in separate systems, the reasoning between them can be lost. A reviewer may see a risk score without its explanation, or a monitoring threshold without a clear record of why it was approved. Reconstructing a past decision becomes a separate investigation.
My responsibility
I developed the product from the underlying problem through architecture, implementation and deployment. My responsibilities covered product ownership, the data model, backend services, merchant and reviewer workflows, security and operations. The platform has been implemented by payment service providers.
The decisions
behind the work.
Make risk decisions explainable
Build deterministic scoring from rules and weights the payment service provider controls. Preserve the answers and policy version behind each score, and separate generated recommendations from approved decisions.
Connect onboarding to monitoring
Turn merchant declarations into a typed, versioned profile. Link that profile to approved controls, deployment records and observed behaviour. Changes to the baseline pass through review, preserving the rationale for each decision.
Preserve the evidence behind the decision
Design point-in-time reconstruction so reviewers can examine the file as it stood on a chosen date. Control and audit access to evidence, isolate tenant data and keep document scanning within an operator-controlled environment. An unavailable check is recorded as unavailable.