KAMWI PAY: A Trust-Bounded and Audit-Evident Architecture for Invite-Native Digital Micro-Lending

KAMWI PAY is an invite-native digital micro-lending platform designed for socially bounded credit formation in Kenya. The system combines identity-gated onboarding, invitation-bounded graph expansion, Bayesian credit-state updates, EigenTrust-style reputation propagation, and an internal tamper-evident ledger. These components are aggregated into a composite lending score that governs borrower visibility, eligibility, and pricing within a bounded trust horizon. The operational layer couples mobile-money collections, bank-led settlement, administrative oversight, and daily Merkle commitments over critical event records. This paper formalizes the scoring model, specifies the ledger construction used in the implementation, and situates the architecture within Kenya's digital credit and data-protection framework. The result is a controlled credit market in which access, reputation, adverse events, and administrative actions remain inspectable.