Stock Brokerage Platform

Nobody on the team, including me, had a finance background; my own academic training is in food science. So the first real challenge wasn't design: it was learning the domain well enough to question it, and becoming the bridge between the client's financial and regulatory knowledge and the development team.

I interviewed admins, advisors and back-office operators to surface pain points, then sat with advisors and back-office teams to watch how they actually worked rather than relying only on what they said. For a period I had direct access to the client's back-office system, which let me learn its workflows first-hand, until that access was withdrawn under compliance rules. To understand the investor side, I opened a CDS account myself. And because unclear requirements in a domain this regulated can't be guessed, I logged open questions directly beside the designs and took them back to the client.

Four insights shaped the platform. Advisors didn't want a new system; they wanted their old system to feel better, after a decade of muscle memory. A signal can change form if it keeps its meaning: buy and sell had to stay instantly recognisable, but the colour didn't have to fill the whole window. Desktop power users judge a web app by what it can't do, so right-click menus, snap-to-grid layouts and saved table settings weren't extras. And access control turned out to be a product in itself.

The admin panel sits at the centre, controlling who can do what in the other three. It covers users, CDS account settings, security and risk settings, announcements, reports, operations and approvals. The back office is organised into 13 sections, from Client and Portfolio to Approvals and Internet Trading, and sensitive actions go through a maker-checker approval step. The OMS is built around a home screen that combines market watch, order entry, the order blotter and the order basket, with market tools such as alerts, heat maps, quotes and trade summaries. The mobile app covers Home, Stocks, Transactions, Reports and Menu, plus CDS account opening.

The core flows run across them: placing an order from the market watch and tracking it in the blotter; building a basket of clients and placing the same order for all of them at once; opening a CDS account on mobile; submitting an action such as a payment or penalty waiver for an authorised approver to accept or reject with a reason; and creating a user in the admin panel, which sets their access everywhere.

Three areas went through the most rounds. The OMS home screen changed repeatedly with advisor feedback, especially the grid settings and right-click options, until every requirement advisors raised was delivered. The CDS account-opening flow was revised with the client's documentation department to meet regulatory needs, then presented to the CSE and approved. And the buy and sell windows and order basket were compared old against new; on my most recent visit, advisors were happy with the new windows and preferred the light theme, which now guides which theme we prioritise.

The constraints were real throughout: strict secrecy, with requirements withheld until NDAs were signed and almost no visuals of the existing system; every process passing regulatory checks with the CDS, SEC and CSE; a six-month timeline extended by another six, still short for four products; and developers and QA engineers leaving mid-project. With days or weeks to design entire flows, I used AI-assisted exploration to generate alternative patterns quickly, then refined them myself.

I led design across all four products, supported by a design intern on back-office screens: I set the layout and structure for each screen, and the intern filled in table details such as column headers and sample data. Around us were two members of senior management, three frontend developers, two backend developers plus one brought in for complex work, two QA engineers and a DevOps engineer, with several developers and QA engineers leaving and new hires joining along the way.

No one on the team had a finance background, so every feature needed a detailed walkthrough. I annotated designs in Figma for the frontend developers, walked them through the concepts behind each screen, and troubleshot with them during build. I worked constantly from CSE documentation, using AI tools to summarise and translate dense regulatory material and explain complex financial concepts in plain terms before passing them on, and I manage the project in Jira alongside the QA lead, using Claude Code to help create tickets.

The project is still in progress, with each product at its own stage: the mobile app (iOS and Android) is fully designed and developed and in client pre-launch testing; the OMS is about 80% developed with client testing underway; the back office is about 80% designed and 10% developed; and the admin panel is about 90% designed with its backend in development. Early results include an estimated 10–15% improvement in workflow efficiency, with early signs it could approach 25% once the platform is fully in use, though a reliable figure will only come from running both systems side by side. Users can now stay signed in on mobile, web and desktop at the same time, CDS account opening on mobile is approved by the CSE, and every advisor requirement for the OMS home screen has been delivered. Because the client is one of the most established brokerages in the country, its move could encourage others to consider an alternative to the incumbent.