Re-architecting payments at $2T scale

The Problem

The Card Line of Business (LoB) was anchored by a legacy monolithic payment application that had become increasingly unstable. This “black box” system was difficult to scale, prone to outages during peak volumes, and acted as a bottleneck for new feature delivery. The technical debt was so significant that even minor changes risked cascading failures across the payment lifecycle.

The Strategy: The Strangler Method

Rather than a high-risk “big bang” migration, we employed the Strangler Fig pattern. We froze the legacy system, limiting its scope strictly to production support and essential maintenance. This ensured stability for existing customers while we systematically “strangled” its functionality by building out a modern replacement in parallel.

Architectural Innovation

We designed a cloud-native, event-driven ecosystem built on Domain-Driven Design (DDD) principles. Key architectural pillars included:

• Backend-for-Frontend (BFF): Implemented using Java/Spring Boot to provide optimized data structures for our React.js frontend, reducing client-side complexity and improving performance.
• Anti-Corruption Layer (ACL): A critical defensive component that translated and sanitized data from the legacy system before it reached the new microservices. This prevented “bad data” and legacy logic from polluting the new architecture.
• Microservices Architecture: Decomposed the monolith into discrete, independently scalable services (e.g., Authorization, Settlement, Ledgering) using Spring Boot and Postgres.

From Application to Platform

What began as a replacement for the Card LoB evolved into a multi-tenant payments platform. We implemented a flexible business rules engine driven by YAML and JSON-based logic. This allowed the platform to support additional lines of business without code changes, enabling them to define their own payment flows, validation rules, and processing logic through configuration.

The Impact

By the end of the transformation, we had successfully migrated ~$2T in annual transaction volume to AWS. The new platform didn’t just improve stability; it transformed the business’s ability to compete. Feature delivery cycles dropped from months to weeks, and the improved throughput and reliability directly contributed to an estimated $30M in incremental annual revenue.