The problem
A lending startup needed to make sense of credit bureau data from three sources — CIBIL, CRIF, and Equifax — across consumer and commercial reports. Each bureau returns deeply nested JSON in a different schema, with different naming conventions, different denormalization patterns, and different ideas of what "account active" means. Credit operations was spending more time normalizing data than evaluating borrowers.
What we built
A bureau ingestion + rendering layer that:
- Ingests structured credit data from CIBIL, CRIF, and Equifax (consumer + commercial variants)
- Normalizes the schemas into a single internal model
- Computes derived insights on top: capital utilization, account tenure, DPD (days past due) bucketing, year-on-year credit trends
- Renders branded, client-ready dashboards that look the same regardless of which bureau the underlying data came from
Status
Full write-up coming soon. This case study covers a recent proof-of-concept project. Get in touch if you want to dig into the architecture before the public write-up lands.