Building for Configurability

I led the strategic transition of ONCI's core platform from a rigid, hard-coded dashboard into a modular, event-driven system. By designing a 'widget-based' architecture and a flexible condition-builder, I empowered banks to configure the platform to their unique credit policies without requiring engineering intervention.

Historically, every new customer at ONCI was a bespoke engineering project. Because every bank has different credit processes and data requirements, onboarding required custom code and significant manual setup from our Credit Intelligence team.

During a period of downsizing, our team was reduced to be more appropriate for our revenue at the time. We no longer had the luxury of dedicated engineers and analysts for every client. We faced a stark choice: we could either onboard new customers OR build the product, but our manual 'monolith' approach prevented us from doing both.

I secured immediate buy-in from senior stakeholders by correlating a new modular architecture directly with the ability to free up restricted resources.

Following the departure of our PMs, I stepped in to manage the Jira backlog, prioritization rituals (via monthly 'Betting Table' sessions), and all sprint ceremonies to keep the team aligned and moving at 2x velocity.

To avoid repeating years of discovery, I used AI to search through legacy company data from SharePoint, Slack, and HubSpot. This allowed me to synthesize deep customer insights and Jobs-to-be-Done without burdening the Sales and CS teams.

The hardest challenge was determining the borderline — what should be user-configurable and what must remain hard-coded to maintain system integrity.

I observed the team getting stuck in circular arguments about which data to display for which user. My research revealed that different departments within the same bank wanted entirely different views of the same borrower.

Instead of designing a 'perfect' page, I designed a modular widget system. This allowed customers to add, edit, and rearrange data components themselves.

To move fast, I designed a rough internal interface for our team to make deep adjustments without code, while focusing my high-fidelity craft on the client-facing UI to ensure it was simple enough for them to adopt without training.

Beyond the UI grid, I designed the 'Alerts' and 'Reports' modules using a highly flexible Condition Builder interface.

Previously, users ignored predictive alerts because they were cumbersome. I designed a lightweight workflow that allowed users to change alert statuses (e.g., New, In Progress, Resolved) with a full audit trail. This transformed a static forecast into a daily risk-management tool.

Empowered users to build their own custom data queries and share them across their organization, removing the need for ONCI analysts to manually generate reports.

By moving from a static, hard-coded dashboard to a fluid, event-driven grid, we effectively 'automated ourselves out of the onboarding loop.'

As a Principal IC, my most valuable contribution wasn't the components themselves, but the stakeholder diplomacy and operational rituals that allowed a leaner team to deliver a far more powerful product than we could in our 'big team' era.