Case Study: LendUp achieves faster feature extraction and daily model iteration with Databricks

LendUp - Customer Case Study

LendUp builds technology to expand access to credit for underbanked and unbanked Americans by using machine learning to evaluate credit risk. Their data science team faced a major bottleneck: feature extraction ran on a single AWS instance and processing millions of semi-structured (XML) documents could take days, while poor integration across storage, modeling, and analytics tools created heavy “glue” work and slow model iteration.

By adopting Databricks, LendUp moved feature extraction to distributed Apache Spark clusters and into an integrated notebook workspace with native S3 access. This eliminated the single-machine bottleneck, cut processing from days to hours, shortened model iteration from weekly to daily, and substantially boosted data scientist productivity—enabling faster, higher-quality credit models that better serve LendUp’s customers.

LendUp

Jacob Rosenberg

CTO, LendUp