REALas is Australia’s most accurate property price predictions service. Predictions are available for over 90% of all properties listed for sale across Australia, available free for the public through our website, and to internal users and external businesses through our API.
I also learned heaps about CI/CD pipelines, agile methodologies, API development and was fortunate enough to have my AWS Cloud Practitioner certification sponsored by the company. Here’s a selection of some of the features I worked on.
A starter task assigned to help me become familiar with the REALas codebase. Converts an address to its corresponding property ID using Google Maps and CoreLogic APIs.
This adds to the tool set available on the internal admin panel used by the developers and support team. The admin panel is powered by Strapi CMS.
Adding social media login integration using Firebase Auth to streamline account creation process for users.
A refactor of the login flow, login API endpoints and database schemas was required to allow for the additional information. However, as logins from all providers are delegated through Firebase and only a Firebase Auth token is stored on REALas’ end, no additional backend code will need to be updated in the future to allow for new social providers.
REALas.com is a Progressive Web App. This feature taps into the service worker API to show a toast notification when a new build has been loaded into storage and is ready for the user to load.
REALas makes tens of thousands of property price predictions every week. Sometimes, usually due to data errors, we make erroneous predictions. The clanger detection system attempts to capture potential errors and flag them for manual review.
As part of this project, I had to develop a lambda to listen to new predictions coming in through an SQS notification stream and determine whether the prediction is likely to be a “clanger”. If a property is likely to be a clanger, this property is flagged on a separate database linked to our admin panel, where I built a full UI to inspect and resolve these predictions.