Harmony

Responsive Data Pipeline

As one of Australasia's leading marketplace lending platforms, Harmoney has matched more than 32,000 borrowers with more than $1 billion in personal loans. Its lending experience is 100 percent online, from application to funding, necessitating complete and accurate data collection, processing, and analysis.

Harmoney's previous data warehouse, implemented in a traditional database solution, could not ingest certain file types or natively connect to a range of cloud-based analytics systems and third-party data services. Its data pipeline used custom-coded Python connectors, which were rigid and hard to maintain.

To speed data-ingestion activities and accommodate new types of data, Harmoney adopted a cloud data platform designed to store vast amounts of data of various types, including relational SQL Server data and a variety of SaaS sources. The platform includes a pipeline that enables Harmoney to ingest 50 times more data than before. Daily loads are 40 times faster and include a variety of structured and semistructured data sources to feed machine learning models. As new data comes in, it automatically runs through the model to make decisions about who to lend money to, in what amount, and at what interest rates, based on perceived credit risks.

Continually ingesting new data allows Harmoney to refresh its sales funnel every 15 minutes and to monitor the funnel via live dashboards, removing friction points for borrowers by offering prompt, accurate loan decisions. Business users now have fast, native access to data ten times faster for business intelligence, predictive analytics, and machine learning activities.