Macquarie Bank — Work — Karlo Estrada

For four years I worked across cross-functional teams at Macquarie, partnering closely with solution design, UX, and engineering. Our squad included a product manager, multiple BAs, a design architect, UI engineers on mobile and web, backend developers, and tech leads. We also ran a design chapter, where designers were delegated across different streams of work: the banking app on mobile and web, adviser online, and loan originations.

My role sat within the consumer banking app, where the initiatives focused on self-service and digitalisation: financial assistance (customers applying for support or a home loan repayment extension), account closures, collections and arrears, account management such as term deposit maturity instructions, and managing user profiles.

A few that stood out

Financial Assistance. Regulation required the bank to clearly communicate available support when customers could not meet repayments. We designed a self-serve flow in the app that let customers apply and receive an outcome, instead of calling and waiting in a queue. The focus was on clear steps, accessible design, and handling errors and edge cases properly. We also worked through policy and scope decisions, such as whether a single application could cover multiple accounts.

Digital TPA. Third party authority lets a customer give another person or service access to their accounts (for example a partner, adviser, or Xero). The process was paper-based and manually processed, and many requests failed due to signature issues. We moved it into the app to reduce rework and make the setup clearer. The core UX problem was making permissions easy to understand: who has access, which accounts they can access, what they can do, and how to revoke access.

Home Loan Facility Map. An uplift, not a new idea. The map shows a home loan as it actually is: a facility, with loans under it, and offset accounts under those, up to ten of them, three levels deep and ordered oldest to newest. A list of rows makes the customer reassemble that structure in their head. Most of the work was in the edges: more than one facility at once, and what the thing does when a level will not load.

Variable Rate Review. An in-app flow for requesting a review of your variable home loan rate, without a phone call. The hard part was scope. A home loan is a facility with several accounts on different rate types, and only some are eligible. The work was making this clear: which accounts are in the review, which are not, and why. We never promised a rate reduction up front, which let the outcome screen be honest about the result.

The work I'm proudest of isn't a screen

As a team, we improved the handoff between design and engineering by integrating AI automation into the business and product requirements for the features we build and maintain.

We introduced a digital product specification: a structured document that explains what a digital product should do, from a product and user-behaviour perspective. It captures the intended user flows, screen behaviour, rules and logic, and the data a user should see, so everyone (product, design, engineering, and stakeholders) shares the same understanding of how the experience should work.

It deliberately avoids backend and system rules, technical build details, and API definitions. Instead it focuses on the decisions that shape the user experience: requirements, edge cases, states, and expected outcomes. In practice, it acts as a single source of truth that keeps work aligned as designs evolve and features move toward delivery.

The core idea is that we already have the product codebase for the actual mobile app. With that in place, AI can read, analyse, and update the codebase to support whatever new feature we want to build.

What enterprise design taught me

Three things stayed with me.

The interesting design problems in regulated environments are rarely on the surface. The screen is a thin layer over an org chart, and if the org chart is wrong, typography will not save it.

Craft survives in enterprise, but only if you fight for it quietly: in pull requests, review comments, and the unglamorous parts of component specs. The loud version gets sanded off.

The most useful thing I learned wasn't a Figma or SwiftUI skill. It was writing design decisions down so they survived a six-month delay and a change of stakeholders. Most design work in a bank doesn't ship the week you finish it, so the brief that travels is the one that matters.

I left Macquarie in 2026 with that process still in the team's hands, and a stronger belief that internal design tooling, the unglamorous kind, is where a lot of the next decade's interesting problems sit.