Case Study · Banking & Financial Services

Rebuilding a global bank's back-office systems while they stayed in production.

A Tier-1 global bank, operating across dozens of countries, needed core back-office systems rebuilt on current technology without pausing the operations that depend on them. rinf.tech joined the bank's internal delivery team to re-architect systems as microservices, move business processes off commercial middleware, and replace email-based case work with an auditable workflow platform, in phases, while the systems kept running.

Five years on, the work spans five divisions, from wealth management to fixed income, and the rebuilt systems are in production.

IndustryBanking & financial services (Tier-1 global institution)
Engagement typeMixed delivery squad, within the client's internal team
DivisionsWealth management, private banking, lending, compliance, fixed income
DurationFive years, ongoing
Executive summary

At a Tier-1 global financial institution, several back-office systems ran on aging commercial middleware. They were central to day-to-day operations across wealth management, private banking, lending, compliance, and fixed income. Freezing them to rebuild was not an option, and replacing them wholesale would have put live operations at risk.

rinf.tech joined the bank's internal delivery team and worked as a mixed squad alongside the bank's own engineers. The work re-architected three of these systems from monolithic applications into microservices, migrated twelve business processes from commercial middleware to an open BPM engine, and connected the rebuilt systems to the bank's online banking channels. A separate workstream replaced email-based case handling with an auditable, role-based workflow platform.

Each change moved in phases, with the legacy path kept available until the new one was confirmed production-ready, so operation stayed continuous. The overhaul of three critical systems lowered running cost and improved performance; the workflow platform cut case response time by about 15% and has run since 2023; and the engagement has grown across five divisions over five years.

12

Business processes migrated
Moved from TIBCO, a commercial middleware platform, to jBPM, an open BPM engine, phased to keep operation continuous.

3

Back-office systems rebuilt
Re-architected from monolithic applications into microservices and connected to the bank's online banking channels.

5

Divisions served
Wealth management, private banking, lending, compliance, and fixed income, over a five-year engagement.

Chapter 1 · The challenge

Systems the bank could not switch off to replace

The bank's back office ran on enterprise systems built over years on commercial middleware. They processed work that touched wealth management, private banking, lending, compliance, and fixed income. Downtime in any of them is felt across the bank the same day.

That is what makes a system like this hard to modernize. It cannot be frozen for a rebuild, because the business runs on it. It cannot be swapped out in one move, because the risk of a cutover failing is too high. And staying on the commercial middleware carried its own cost: licensing, and logic locked inside a platform the bank did not control.

Alongside those systems, much of the bank's case work still moved through email. There was no shared record of who owned a case, what had already happened on it, or whether the person a thread was forwarded to was allowed to see the client information inside it. For a regulated institution, that was its own exposure.

Agreed before development
  • Operation stays continuous. Processes keep running through the migration. There is no freeze window and no outage planned into the rollout.
  • Move off commercial middleware. Business logic moves from the proprietary platform onto an open BPM engine, removing licensing cost and vendor lock-in.
  • Case work becomes auditable. Email-based handling moves to role-based, access-controlled processing with a shared audit trail.
  • Integrate with online banking. The rebuilt systems connect to the channels the bank's customers already use, rather than running to one side.
  • Each change moves with a fallback. Nothing goes live without the legacy path still in place, so a problem never takes a live operation down.
Chapter 2 · What rinf.tech helped develop

A migration shaped to keep the bank running

The approach followed from the constraint that nothing could stop. Rather than rewrite each system in one pass, the work decomposed the monolithic applications into microservices, each owning a bounded part of a process, and moved the processes themselves onto engines the bank could govern directly.

How a process moved

# 01 — Model the process in BPMN

Each legacy process was mapped as an explicit, versioned BPMN model, rather than left as logic buried in middleware configuration. The model became the record of how the process actually runs.

# 02 — Rebuild as microservices

The monolithic system was decomposed into services running on OpenShift, each responsible for one part of the process and communicating over REST APIs, so a change in one area no longer meant redeploying the whole system.

# 03 — Migrate off commercial middleware

The process moved from TIBCO to jBPM, an open BPM engine. The migration ran in phases, with the legacy path kept available until the new one was confirmed production-ready.

# 04 — Integrate and cut over

The rebuilt process was connected to the bank's online banking channels and data, then cut over once it ran cleanly in production, with the legacy fallback retired only afterward.

The systems were built on Java, with Spring Boot and Spring Data for the services, RxJava for asynchronous processing, and MapStruct for data mapping between layers. Services communicated over REST APIs on OpenShift. Elasticsearch handled search across records, Oracle Database held the data, and the user-facing applications were built in Angular.

Replacing email case work with an auditable platform

A separate workstream took on the case handling that still ran on email. It replaced that with a platform on Camunda, an open-source BPM engine: a case enters through its workflow, the platform checks the user's role and clients against the bank's entitlement data before any task is visible, the case routes automatically between role-based queues, and on completion it archives to the bank's records system with its audit trail. It rolled out in phases, with the email process kept as a fallback until each workflow was confirmed production-ready.

Built inside the bank's own team

rinf.tech worked as a mixed squad within the bank's internal delivery team, not as an isolated external build, and took product and project leadership in several areas of the work. The phased rollout ran the same way each time: a defined scope moved first, the rest kept working the existing path, and the legacy fallback stayed in place until the new process was confirmed in production.

Chapter 3 · Results

What the overhaul delivered

Three critical systems rebuilt on current technology.

Overhauled from monolithic applications into microservices, in production, and well received by the divisions that depend on them.

Lower running cost.

Moving twelve processes off commercial middleware removed its licensing and the overhead of maintaining logic inside a platform the bank did not control.

Better performance.

Microservices on OpenShift replaced systems whose every change had meant touching the whole application, so parts of a process can now scale and be updated on their own.

Case work off email, onto an auditable record.

The Camunda workflow platform replaced email-based case handling with role-based, auditable processing. Case response time improved by about 15%, and the platform has run in production since 2023.

Continuous operation through the migration.

Each change moved in phases with the legacy path in place, so live work was not interrupted while the systems were rebuilt.

The engagement has run for five years and continues, growing from its first systems into work across five divisions, from wealth management to fixed income.
What it proved

Modernizing without an outage window

This engagement shows what it takes to rebuild mission-critical systems at a regulated institution without stopping them: processes modeled explicitly in BPMN, monolithic applications decomposed into microservices, email case work moved onto an auditable platform, a move off commercial middleware onto engines the bank governs, and a phased rollout with a legacy fallback at each step.

For banks carrying core processes on aging middleware, this is what the alternative to a freeze-and-replace looks like.

Industry
Banking & financial services (Tier-1 global institution)
Client
Anonymized at its request
Engagement type
Mixed delivery squad, within the client's internal team
Scope
Legacy-to-microservices re-architecture, BPM migration (TIBCO to jBPM), auditable workflow platform (Camunda), online-banking integration, QA
Divisions
Wealth management, private banking, lending, compliance, fixed income
Duration
Five years, ongoing