Reinventing mobile banking with ML

Overview

HSBC used Azure Databricks and Delta Lake to unify data analytics across data engineering, data science and analysts for its PayMe mobile payments app, applying NLP and machine learning for fraud detection, personalization and recommendations. Data processing for complex analytics dropped from 6 hours to 6 seconds, 14 read replica databases were consolidated into one Delta Lake, and app engagement improved 4.5x.

The challenge

HSBC struggled to overcome scalability limitations that blocked it from making data-driven decisions for its PayMe mobile payments app: legacy systems left data science teams working with data that was weeks old after manual, time-consuming exports; every data request required a manual, error-prone approval and masking process; data scientists worked in silos on custom environments, limiting collaboration and slowing model iteration; and data analysts struggled to get the structured data subsets they needed for business intelligence and reporting.

The solution

With Azure Databricks, HSBC unified data analytics across data engineering, data science and analysts, using NLP and machine learning to understand the intent behind each PayMe transaction and power recommendations, personalization and fraud detection. Delta Lake enabled real-time, secure anonymized data masking for data science and analyst teams and gave HSBC performant, scalable pipelines for downstream analytics and ML, while auto-scaling clusters improved operational efficiency across the machine learning lifecycle.

Reported business value

Complex analytics processing time dropped from 6 hours to 6 seconds, 14 read replica databases were consolidated into a single Delta Lake, PayMe became the #1 app in Hong Kong with 60% market share, and network science-driven personalization improved app engagement 4.5x.