As an ML Engineer at REA, I led technical R&D for GenAI experiments on our flagship realestate.com.au and flatmates.com.au sites. Engaged as a technical conduit between product/marketing stakeholders and dev teams, I technically advised in brainstorming sessions, led comprehensive feasibility research in the use cases considered, performed cost assessments (ensuring that live experiments operated within budget), and devised guardrails to developers who implemented experiments in their custodian codebases.

Additionally, I led the development of a property attribute extraction ML pipeline from property listing images. Performing comprehensive technology analysis, pricing estimates, performance evaluations, pipeline implementation, and model optimisation, I ensured the project could be scaled out to handle millions of images at once via rigorous performance testing. This initiative resulted in the successful extraction of new attributes through ML at scale from unstructured data sources, hydrating new property features to hundreds of thousands of property listings each month.

Beyond technical contributions, I fostered a strong engineering-first culture and knowledge collaboration within the team, including a revamp of our team’s documentation processes and producing artefacts such as overarching model architecture and data-flow diagrams.

Outside of my core role, I helped elevate AI literacy across REA. Tailoring numerous presentations on (Gen)AI to diverse (non-)technical audiences, this education uplift garnered praise for making AI more accessible to all by many. Notably, my company-wide AI literacy talk for REA's 2023 AI-themed hackathon reached 450+ individuals.

In June 2023, I had the privilege of representing REA at NewsCorp Global Technology’s inaugural AI Hackathon in Princeton NJ, where I presented REA's innovative approach to AI and ML to an audience of over 100+ (including Google, GitHub, and AWS) that was also exceptionally well-received.

Uplifting realestate.com.au's personalisation stack with real-time streaming.

Curate, cleanse, analyse, and visualise (un)structured data to produce experiment datasets and provide insight to clients, expressing insights in research papers.

Developed new techniques to robustly integrate cloud-based computer vision models. See http://hdl.handle.net/10536/DRO/DU:30152816.