Founded in 2007, the customer has become the leading online marketplace for small business funding, having arranged more than $1.6 billion in funding for thousands of companies throughout the US. The company's mission is to provide small businesses with the best lending solutions in a flexible and transparent environment. The client's proprietary platform matches borrowers to sources of capital based on each company's unique profile - in a safe, efficient, price-transparent environment.
PRODUCTS & SERVICES
The client needed a solution that would leverage its site activity and other data points to predict charge-offs, which is debt that is unlikely to be collected, so it could make informed business decisions.
After a deep engagement with the client to define their business objectives and challenges, SNP proposed a production architecture that included:
- Azure Machine Learning (Cortana Intelligence Suite) to predict charge-off from site data.
- Develop file share for uploading site visit data in CSV or Excel format extracted from electronic PDFs and inspection pictures.
- Schedule job to move data from file share to blob storage at a pre-defined time.
- Generate initial tag dictionary to store unique tags for processing historical images.
- Create a business rules engine - an interface for creating new business rules.
- Create an Azure learning API model for testing.
- Improved reliability of image tagging.
- The client can now analyze diverse data sets quickly to make informed data-driven business decisions.
- Able to quickly and easily uncover hidden opportunities by analyzing all data, whether streaming or at rest.