The workflow before
Abound's data sat in core systems that were hard to extract from, with older records on optical storage. Analysts could reach about 6 months of history. Governance rules for member data were not defined, so there was no trusted base for soft-pull credit scoring or a single view of each member.
What we built
A lean team of a senior data engineer, an architect and an analyst built automated pipelines that pull data from Abound's source systems into a central Azure data lake, ending the reliance on optical storage.
We scoped the governance rules for member data, covering security and accuracy, and modeled the data in Snowflake so analysts can query years of history instead of months. The work ran over 3 months.
Before and after
| Measure | Before | After |
|---|---|---|
| Member history available | 6 months | About 5 years |
| Data extraction from source systems | Manual, partial | Automated pipelines |
| Archive access | Optical storage | Central data lake |
What's next
With history and governance in place, Abound can build its soft-pull credit scoring model and member 360 view. That is the L2 step: models and assistants that work directly on governed member data.
