At Zafin, our mission is to assist banks modernize their core infrastructure to ship distinctive, customized experiences to their clients. To find out buyer relationship tiers and supply customized rewards, we would have liked to course of vital information volumes—100 million accounts with over 10 billion balances. Historically, we’d have carried out this tiering utility utilizing an open-source framework, Java Spring Boot, with a devoted PostgreSQL server. Nonetheless, constructing and scaling this answer with a conventional tech stack proved inefficient and unsustainable at scale.
Conventional information infrastructure doesn’t scale
Like many organizations in monetary providers, we initially relied on devoted information warehousing options to energy our analytics. Whereas useful, this strategy got here with vital challenges: excessive compute prices, upkeep overhead, and efficiency bottlenecks.
With a single-tenant structure, each buyer deployment requires devoted infrastructure. Since some clients use analytics sporadically whereas others depend on it actively, this setup was costly and inefficient. Devoted infrastructure demanded fixed fine-tuning and incurred fastened prices, no matter utilization—a big effort for our engineering groups. Scaling to accommodate large information volumes throughout thousands and thousands of accounts, balances, and transactions typically stretched our conventional methods to their limits.
Advanced computations, corresponding to processing billions of data for buyer segmentation, tiering, and analyzing relationship patterns, ran as batch jobs that have been gradual to finish. These bottlenecks delayed time-to-insight, and our conventional strategy couldn’t sustain with the dimensions and pace banks required.
To seamlessly present superior analytics to our clients, we would have liked an economical, scalable platform able to dealing with large information volumes, delivering excessive efficiency, and retaining prices underneath management.
Migrating to Databricks to speed up information intelligence
We ran a proof of idea (POC) to match our earlier Java Spring Boot/PostgreSQL setup with the Databricks Information Intelligence Platform. Right here’s how the POC empowered us to enhance our information infrastructure.
Databricks delivered as much as 10x quicker computation speeds in comparison with our earlier answer. For instance, duties that beforehand took 4,000 seconds on our earlier system are actually accomplished in simply 300 seconds. With elastic scalability, we are able to course of billions of transactions and balances throughout thousands and thousands of accounts with out compromising efficiency.
Migrating analytics workflows diminished time-to-market for brand spanking new capabilities by 30–40% whereas requiring fewer engineering assets. In comparison with our conventional setup, Databricks helped us speed up the supply timeline of our tiering utility by 30%. With much less effort spent on infrastructure administration, our engineering group can concentrate on constructing progressive options for our clients.
Databricks’ Serverless capabilities have been a sport changer, permitting us to separate compute from storage to make sure we solely paid for the compute we would have liked. This strategy eradicated the fee burden of idle infrastructure. Cloud prices for analytics dropped by 50–70%, relying on utilization profiles. The power to auto-scale based mostly on information masses meant that each high-traffic and sporadic-use clients might depend on a seamless, responsive expertise with out guide tuning.
The outcomes have been clear: Databricks supplied a transformative answer that delivered superior efficiency, scalability, and cost-efficiency—all whereas assembly the stringent governance necessities of extremely regulated monetary establishments.
Delivering worth to our clients
Databricks’ unified setting permits our groups to deal with information ingestion, transformation, and analytics in a single place—enhancing productiveness and collaboration. From information ingestion to complicated SQL-based transformations and embedded analytics, all the pieces runs seamlessly on Databricks the place even non-technical customers can derive significant insights. When talking with Terry Hickey, Zafin’s Chief Income Officer, he expressed, “With the modernization of our information infrastructure, we are able to now assist banks effortlessly uncover actionable insights to drive progress and deepen buyer relationships.”
With Databricks because the spine of our trendy information stack, we’ve been in a position to innovate quicker and ship better worth to our clients, together with:
- Personalization: Utilizing Databricks SQL and Notebooks, we constructed a data-intensive tiering utility that evaluates buyer balances, product holdings, and transaction patterns to find out relationship tiers (e.g., Gold, Silver, Platinum). This permits banks to ship customized rewards and advantages to clients whereas nudging clients towards the following tier with focused engagement, rising retention and loyalty.
- Enterprise Intelligence (BI): Our embedded analytics answer, Zafin Analytics, helps line-of-business bankers achieve actionable insights into product efficiency and provide effectiveness. With Databricks’ built-in setting, we offer out-of-the-box analytics to reply vital questions like: How is a checking account performing out there? What’s the adoption fee of a promotional provide? How do buyer transaction patterns differ throughout tiers?
- Price-Environment friendly Transformations: With Databricks Autoloader, we seamlessly ingest billions of data from varied file sorts (Avro, CSV) saved in Azure Blob Storage. This, mixed with Serverless scaling, ensures that we are able to course of information effectively, with out over-provisioning or incurring pointless prices.
- Making certain Information Governance and Compliance: Unity Catalog permits sturdy information governance, permitting us to isolate single-tenant buyer environments and keep compliance with monetary providers rules.
The influence for our banking clients is profound. The identical tiering utility that beforehand took a big financial institution two years to develop can now be deployed by their clients in simply months. It is a vital time-to-market benefit that drives quicker innovation and monetization.
“It’s thrilling to see how Zafin is leveraging the Databricks platform to ship information and AI-driven improvements that empower banks worldwide. By modernizing their infrastructure, Zafin is enabling monetary establishments to unlock new alternatives for personalization, buyer engagement, and growth-faster and extra effectively than ever earlier than.”
— Junta Nakai, Vice President – World Head of Monetary Companies, Cybersecurity and Public Sector GTM | Databricks
Wanting forward: Tapping into real-time improvements
Along with enabling quicker, more cost effective information processing, the Databricks Platform has positioned us for future innovation. We’re exploring alternatives in AI and GenAI, corresponding to real-time tiering insights, profitability modeling, and real-time affords with Databricks’ Delta Dwell Tables and streaming capabilities. To additional streamline our analytics capabilities, we additionally plan to allow Lakehouse Federation for seamless integration throughout information sources.
As we proceed to scale and innovate, we’re excited to unlock much more highly effective capabilities for our clients—reworking information into actionable insights that drive the way forward for banking.
About Zafin
Based in 2002, Zafin is a world supplier of SaaS options for core banking modernization and transformation. Our award-winning platform permits banks to innovate their enterprise fashions whereas modernizing their know-how, guaranteeing transparency and equity for banks and their clients. By enhancing operational effectivity, boosting income, and enhancing buyer experiences, Zafin empowers monetary establishments to realize their strategic objectives.
Zafin is headquartered in Vancouver, Canada, with a presence spanning places of work and clients worldwide, together with main banks like ING, CIBC, HSBC, Wells Fargo, Navy Federal Credit score Union, PNC, and ANZ.
Study extra
Executives and Enterprise Leaders can achieve insights from friends on the Monetary Companies Discussion board–Shifting to Monetary Intelligence, in New York Metropolis on Mar 20, 2025. Register right now.
For technical professionals, go to the Information Intelligence for Monetary Companies webpage for extra use instances and assets.
