Аннотация
With the progress of digitalisation in recent years, many new risks have appeared for second-tier banks in Kazakhstan, such as cybercrime, weak authentication, liquidity problems caused by high-frequency trading, etc. Given the problems mentioned above, a hierarchical suspicious transaction detection mechanism based on recursive least squares and clustering has been introduced; a distributed ledger will be employed to guarantee the immutability of transactions, and machine learning dynamic thresholds will be used to deal with the heterogeneity of behaviour. Given the present problems in digital resilience, this paper will present the concepts of network segmentation and micro-segmentation in a zero-trust model, propose a plan for migration to post-quantum cryptography, and put forward a convergence model for anti-money laundering rules and smart contracts in cross-border payments.