Selected Work
A focused portfolio of machine learning systems, cross-platform applications, and data engineering projects — each delivered end to end.
SmartType 4 — Deep Learning Engine
Delivered the core deep learning chromosome classification engine for Digital Scientific UK Ltd's next-generation karyotyping platform. Achieved 97%+ classification accuracy across all 24 chromosome classes, alongside bespoke medial axis generation for chromosome orientation analysis and a full CNN training pipeline with ONNX cross-platform deployment.
Smart Trainer — ML Training Interface
Bespoke desktop application for training and monitoring EfficientNet-based deep learning classifiers. Features real-time 60fps animated feature map visualisation, drag-and-drop dataset validation, and hardware-aware rendering that adapts dynamically to available GPU and CPU resources. Designed to enable non-technical users to run model training independently.
GROWL — Cross-Platform Voice Application
End-to-end encrypted push-to-talk application built with Flutter, targeting Android, iOS, and Linux desktop from a single codebase. Implements peer-to-peer audio via WebRTC with Opus codec, Bluetooth LE and NFC device discovery, and a custom WebSocket signalling server — delivered through 27 iterative sprint releases.
CytoFetch — Genomic Data Pipeline
Automated extraction and normalisation pipeline bridging UCSC and NCBI genomic databases with clinical software requirements. Delivers full ISCN coordinate normalisation across multiple species, embedded data provenance tracking, and validated output for direct integration with SmartType 4.