-
Collected and prepared a large dataset of valid and fraudulent historical transactions. This was used to train the AI system. The data was prepared for training by utilizing pre-processing techniques such as data cleaning, data normalization and feature engineering.
-
Machine learning algorithms were utilized to analyze the transaction data and behavioral data, and identify patterns linked to fraudulent activity
-
The AI-based solution was then integrated into the current system of the client and transactions were screened in real-time.
-
The system was monitored continuously and used techniques like A/B testing to adjust to new patterns of fraudulent activity.
-
AML (Anti-Money Laundering) checks and monitoring systems (such as solutions for customer identification and verification, transaction monitoring, reporting suspicious activity to stakeholders, etc.) were implemented to detect and prevent money laundering activities. The system alerts the AML team of suspicious transactions that need to be investigated. Collaboration with the team is required to establish alert thresholds and assess system performance.
-
A chatbot was built using natural language processing (NLP) and machine learning technologies to aid with compliance (KYCs) and fraud detection, and to assist customers. The chatbot was integrated with the AML system and was trained to detect phishing scams and other suspicious activities. Other than that it was trained to assist customers with common queries such as account balance inquiries, transaction history, etc.
