Proud to spotlight our partnership with Databricks and how it’s transforming the way we deliver merchant risk intelligence.
The challenge
Rapid growth brought us rapid challenges. Siloed data, limited visibility, and onboarding timelines stretching for weeks. As we scaled across new geographies and payment providers, the cracks in our infrastructure became harder to ignore. We needed a platform that could keep pace with the volume and complexity of the data flowing through our systems.
What we built
By unifying our data and AI workflows on the Databricks platform, we fundamentally changed how we operate:
- Automated ingestion from 30+ payment providers. Data that previously required manual processing now flows directly into our lakehouse, standardised and ready for analysis.
- Cut merchant onboarding to a single day. What used to take weeks of manual underwriting and data gathering is now handled by automated scoring pipelines running on Databricks.
- Real-time insights for teams and customers. Both our internal risk team and our customers now have access to live dashboards and alerts powered by the same unified data layer.
The architecture
Our implementation follows a medallion architecture pattern: raw data lands in the bronze layer from payment providers, gets cleaned and standardised in silver, and feeds our risk scoring models and customer-facing dashboards from gold. The entire pipeline runs on Delta Lake, giving us the reliability and performance we need at scale.
The AI and ML workloads run natively on Databricks too. Our risk scoring models, which analyse 30+ data sources per merchant, train and serve predictions directly from the platform. No separate ML infrastructure to manage.
Where we are now
We’re operating across Asia, Europe, and Latin America with a fully standardised and scalable infrastructure. The partnership with Databricks has been foundational to getting here, not just as a technology platform but as a team that understood what we were trying to build and helped us get there faster.
Read the full case study on the Databricks website: Envisso on Databricks