Money laundering is a substantial danger to financial systems, with money mules playing an important role in hiding the origins of illicit funds. The current paper presents a novel hybrid framework for detecting money mule accounts within India’s Unified Payments Interface (UPI). We integrate domain-driven heuristics with probabilistic modelling via Markov Chain supported by comparisons with state-of-the-art machine learning models. Our approach captures temporal patterns in account behavior, leading to better identification of money laundering strategies. The Markov chain model outperforms other techniques, making it suitable for Anti-Money Laundering (AML) applications. This work advances the state-of-the-art in AML by contributing an interpretable and scalable detection methodology.