Compliance Isn't Slowing Your Onboarding; Your Architecture Is.

Digital payments crossed $18.7 trillion in global spend in 2024, an almost elevenfold increase from a decade earlier, and Worldpay projects the figure will exceed $33.5 trillion by 2030. McKinsey's 2025 Global Payments Report puts the industry's revenue at $2.5 trillion on roughly $2.0 quadrillion in value flows, supported by 3.6 trillion transactions worldwide. Yet for all that scale, the front door of fintech remains stubbornly narrow. Fenergo's 2025 industry survey found that 70% of financial firms lost clients in the past year due to too slow or fragmented onboarding, with average client abandonment now hovering around 10%. The gap between aggregate growth and individual user friction has become the defining engineering problem of the decade.

Ritvik Pandya, a distributed systems engineering leader and Senior IEEE Member with more than seventeen years of experience designing enterprise payment platforms, sees the challenge as fundamentally architectural. His work spans real-time payment infrastructure, cloud-native microservices, service mesh architectures, event-driven workflows, and merchant onboarding systems designed to support financial products across markets and product lines. His perspective on why traditional onboarding architectures fail under modern fintech demands begins with a structural observation: the systems were not built for the ambitions placed on them.

When Tight Coupling Becomes a Tax on Growth

Roughly 79% of IT leaders now say legacy systems are directly limiting their ability to execute digital transformation. The cost of those constraints lands hardest on banks and payment providers. Average annual spend on AML and KYC operations stands at $72.9 million per institution, and UK corporate banks report client onboarding cycles averaging more than six weeks. About 85% of enterprises run on microservices today, but the migration is uneven, and many large financial platforms still operate hybrid stacks where legacy components hold the rest of the system back.

In his work on enterprise payment platforms, Pandya leads the architecture and delivery of a merchant services onboarding platform built around workflow-driven, multi-step journeys that cover KYC and CIP compliance, credit review, underwriting, and partner integration, rather than the static, tightly coupled flows imposed by legacy systems. The platform supports a strategic push to expand merchant services beyond enterprise clients into SMB and ISV segments, where speed of onboarding and configurability of workflow are the deciding factors. Each step is designed as a composable service rather than a hard-wired stage, which lets the platform absorb new product requirements without rewriting the spine. The shift is less about introducing new technology and more about isolating failure domains so that one slow downstream system does not stall the entire pipeline.

"When you trace a stuck onboarding to its root cause, it's rarely the work itself that's hard. It's the assumptions baked into how services call each other," Pandya explains. "A synchronous chain of fifteen services means any one of them can stop the whole thing, and that's not architecture, that's coincidence."

Compliance Across Borders Is an Architecture Problem

Global AML enforcement totaled $3.8 billion in penalties in 2025, and Fenergo data shows more than half of corporate and institutional banks now spend between $1,500 and $3,000 to complete a single client KYC review. Average KYC abandonment rates sit between 25% and 40% across fintech, driven by document friction, slow checks, and inconsistent local rules. Cross-border expansion compounds the problem because every jurisdiction layers its own identity verification standards, sanctions screening, and beneficial ownership rules on top of the existing pipeline. For a platform pushing into new markets, compliance is not a back-office concern but a primary determinant of how quickly the business can grow.

The platform Pandya leads was designed with this reality in mind. The merchant onboarding portals run across multiple geographies, including the US and Canada, with locale-specific compliance, multi-language support, and configurable workflows that adapt to each region's regulatory profile. A dedicated credit review service and underwriting tool sit inside the same event-driven flow as the front-end onboarding journey, integrating with enterprise data systems so that credit ops and risk decisions move in step with the customer rather than behind them. Compliance is treated as a participant in the workflow, not a gate placed in front of it.

"Compliance teams aren't slowing things down. The architecture is. If your only design pattern is to wait for the regulator's check before doing anything else, you've already chosen latency over scale," Pandya observes. "The right move is to design for compliance to happen alongside the work, not in front of it."

Why Event-Driven Processing Has Become the Default

About 85% of global enterprises now use event-driven architecture, and more than 150,000 organizations rely on Apache Kafka as the backbone for real-time data flow. The event-driven architecture software market reached $8.63 billion in 2024 and continues to expand. Major payment platforms have leaned into the pattern: Shopify processes up to 66 million Kafka messages per second across checkout, payments, and inventory, while SumUp, in 2025, used Confluent Cloud to handle millions of payment events daily across more than thirty countries. The shift from synchronous request-response to asynchronous event streams is no longer experimental; it is the operating model for any platform expecting to scale across geographies and product lines.

Pandya's onboarding platform reflects that consensus. A Backend-for-Frontend orchestration layer built on Kafka decouples the merchant-facing UI from downstream systems, including credit ops platforms, underwriting engines, document stores, and partner banks, and pushes real-time updates across the chain rather than blocking on each handoff. He has remained engaged with the broader engineering community through his membership in the IEEE Computer Society. The architecture's choices reflect that depth: event-driven workflows, partitioned consumers, and observability layers that treat backpressure and replay as normal operating conditions rather than exceptions.

"Event-driven design isn't a buzzword. It's a mathematical concession to physics," Pandya notes. "Once you accept that downstream systems will fail, slow down, or change shape, you stop pretending you can hold them on a synchronous string."

Reusable Platforms and the Partner Ecosystem

Marketplace and partner integration have become a structural requirement rather than a nice-to-have. Shopify alone generated $378 billion in gross merchandise value in 2025 across roughly 4.8 million merchants in 175 countries, and the global e-commerce market is on track to reach $4.8 trillion in 2025. Embedded finance, where financial products are delivered inside non-financial software, was valued at $104.8 billion in 2024 and is projected to grow to $7.2 trillion by 2030, according to Dealroom and ABN AMRO Ventures. Every one of those numbers represents a commercial relationship that has to be onboarded, verified, activated, and maintained, often across multiple jurisdictions and partner stacks at once.

In his work on enterprise banking platforms, Pandya led the development of a Micro Frontend platform now adopted across multiple banking and commerce portals, letting each surface plug into a shared, independently deployable set of onboarding components rather than rebuild from scratch. He also stood up a Client Test Environment from the ground up, which enabled external partner banks to integrate against the platform end-to-end, and ran the delivery as part of a multi-region program with hundreds of engineers across product, infrastructure, and credit ops teams.

"Each partner bank looks like its own product. Each portal thinks of itself as its own product. The architect's job is to find the abstraction that lets all of them ship without arguing about it," Pandya, a member of the IEEE Technology and Engineering Management Society says. "Get that wrong and every new partner becomes a quarter of engineering work."

Crypto, Wallets, and the Next Wave of Onboarding

Digital wallet adoption reached 5.6 billion users globally in 2025, covering roughly two-thirds of the world's population. Cryptocurrency has crossed 560 million users worldwide, and the embedded finance market is on track to reach $7.2 trillion by 2030. New product categories, including real-time account-to-account payment rails, stablecoin settlement, and embedded credit, all expect the same fluid onboarding experience that consumer apps have trained users to demand. The platforms that win the next decade will be the ones whose onboarding architecture treats each new product as a native flow rather than a bolt-on.

Pandya’s work applies the same architectural principle to broader payment platform modernization. As financial institutions expand across products, partners, and regions, their systems must support new requirements without forcing teams to rebuild the core platform each time. In onboarding, that means treating each new market or product as a configurable workflow rather than a custom engineering project.

For Pandya, that shift is not only about speed. It is about designing systems where speed, compliance, resilience, and auditability can coexist. Eventual consistency cannot remain a purely technical choice buried inside the engineering team; in financial platforms, it has to be understood across product, risk, operations, and compliance so that everyone has a shared view of what it means for a customer, application, or partner to be “in progress,” “approved,” or “activated.” Observability becomes the trust layer that lets teams see how work is moving across the platform, while retry logic, replay mechanisms, and dead-letter queues become more than engineering safeguards. In regulated financial systems, they are part of the evidence trail showing how the platform handled exceptions, delays, and recovery.

That is why the hardest part of modernization is often not adopting new tools, but changing the assumptions built into the workflow. A platform can use Kafka, microservices, or cloud-native infrastructure and still behave like a legacy system if every process is designed as a blocking sequence. The real shift happens when onboarding is treated as a living platform: observable, configurable, resilient, and designed for the reality that financial products, compliance requirements, and partner ecosystems will continue to change.

"The next financial product won't wait for a six-week onboarding cycle. It will assume real time and forgive nothing," Pandya concludes. "The companies that have done the architectural work won't have to think about it. The ones that haven't will keep losing customers to the door they built themselves."