Seven years shipping production mobile apps. Now building products that think, with ML, LLMs, and the instinct to ship that only comes from doing it for real._
Multi-agent LLM web app analysing Sydney liveability by suburb. Combined Reddit sentiment analysis using DeBERTa and GoEmotions, OpenStreetMap, ABS, and TfNSW transport data into a unified assistant. Led the full architecture for a team of five. Won the Best AI LLM Project Award at UTS 2026.
Interactive data narrative examining the growing tension between housing affordability and public transport access across Australia's 8 capital cities from 2020 to 2025. Built for policy makers using real data from ABS and BITRE.
Exploratory text analysis and data storytelling project combining topic modelling, sentiment analysis, and visualisation into a cohesive markdown narrative report. Graded H with 90/100.
PyTorch image captioning model using a CNN encoder and LSTM decoder, trained and evaluated on the VizWiz dataset (7,750 images from people who are blind). Two architectures designed and compared using BLEU-1/2/3/4 metrics.
Comparative study of CNN transfer learning strategies on 101-class food image classification. Covered head replacement, frozen backbone training, and layer-by-layer fine-tuning across GoogLeNet, MobileNetV3, and ResNet50.
Implemented a Perceptron with forward and backpropagation using only NumPy. Trained a custom neural network on the Japanese MNIST dataset (70k Hiragana images, 10 classes) achieving over 80% accuracy using PyTorch.
End-to-end ML data product predicting next-day high prices for Bitcoin, Ethereum, XRP, and Solana. Individually responsible for the Ethereum prediction model and the data pipeline and news feed powered by the CoinGecko API. Predictions served via FastAPI and visualised in a Streamlit dashboard.
Machine learning service predicting rain in 7 days (binary classification) and cumulative precipitation in 3 days (regression) for Sydney, using Open-Meteo historical data. Deployed as a containerised REST API on Render with Docker.
Interactive geospatial platform developed for the Ministry of Agriculture of Honduras, giving users access to maps and data about agri-food sector production across the country. Built end-to-end from backend to frontend.
Led the complete rebuild of the Ryte iOS application in Swift 5, achieving a 90% reduction in client-reported bugs and a 40% improvement in performance. Included a full redesign of UX and integration of real-time features.
iOS application built from scratch in Swift 5, from mockup design in Sketch to App Store release. Used by 20+ enterprise clients to train employees in safety and security practices.
Open source Swift UI framework published on CocoaPods, providing a customisable alternative to UIAlertController for displaying messages and custom alerts in iOS applications.
Open to AI engineering and full-stack roles where production instincts matter. Remote, hybrid, or Sydney in person.
Get in touch