How I build things that matter

ERA's original agent portal was built on Java JSP with a Laravel backend — a robust but ageing architecture that required significant maintenance effort. The platform handles property submissions, agent dashboards, commission tracking, and back-office operations for ERA's entire Singapore network. The challenge was maintaining and extending a complex hybrid system while keeping 10,000+ agents productive with zero downtime.

Maintained and extended the JSP frontend while progressively modernising backend logic through Laravel. Optimised Oracle database queries for large-scale transactional workloads, integrated jQuery-driven UI enhancements for improved agent UX, and managed deployments on Windows EC2 instances on AWS.

As ERA's agent network grew and expectations for modern UX increased, the JSP-based portal needed a full modernisation. The goal was to rebuild the agent platform as a responsive, fast, and maintainable Nuxt.js application — without disrupting existing operations — while migrating the same Oracle backend data layer and adding Cloudflare for security and performance.

Led the full-stack rebuild using Nuxt.js for the frontend with Tailwind CSS and SCSS, backed by a Laravel API layer, and Oracle for data persistence. Integrated Cloudflare for DDoS protection, CDN caching, and SSL termination. Designed a component-driven UI that gave agents a dramatically faster, more intuitive experience compared to the legacy JSP portal.

Commission submissions and property transaction approvals at ERA were processed manually via paper forms and spreadsheets — creating a bottleneck of hundreds of submissions per day. Errors, missing documents, and approval delays were routine, costing agents income and the business significant back-office overhead.

Designed and built a fully digital submission pipeline: structured online forms with real-time validation, document upload with automated checks, multi-level digital approval workflows, and a commission calculation engine that replaced all manual spreadsheet processing.

ERA's training arm — RIA School of Real Estate — needed a complete digital platform covering two distinct audiences: students (agents) who browse, register, and pay for courses; and admin staff who manage course creation, class scheduling, payment reports, and PayNow refunds. Both portals needed to operate against a shared Oracle database hosted on AWS EC2 Windows servers, with no existing digital infrastructure to build on.

Built two separate portals with distinct tech stacks suited to each use case. The student-facing portal (riaschool.com.sg) was built with Laravel, jQuery, and CSS — featuring a full e-commerce-style course flow: browse, add to cart, automatic promotions, checkout, and registration. The admin portal (admin.riaschool.com.sg) was built with Java JSP, jQuery, and CSS — giving staff tools to create courses, manage class schedules, process PayNow payments and refunds, and generate payment and attendance reports. Both share a single Oracle database on AWS EC2 Windows.

Real estate agents across Southeast Asia spend hours daily on repetitive content tasks — writing property listings, drafting client emails, creating marketing materials, translating documents, and preparing presentations. ERA needed a platform that would bring AI productivity tools directly to agents, without requiring any AI knowledge — and that could scale across multiple countries with a sustainable credit-based business model.

Co-built ERA AI LAB as a full multi-feature SaaS platform powered by OpenAI. The platform provides two core modes: Guided Creation (9+ structured AI tools for specific real estate tasks) and Advanced (a free-form AI chat interface with real estate prompt examples). Built with Nuxt.js, Vue.js, and a Laravel API backend — with a complete user management system, credit-based usage model, multi-country approval workflow, and an admin reporting dashboard.