Mapping structure into unstructured spatial data — orchard canopies, heritage ruins, multimodal animal scans — one point cloud at a time. MSc candidate in Artificial Intelligence, seeking a PhD in 3D computer vision.
My work sits at the intersection of 3D geometry and learned perception — teaching models to find reliable structure in noisy spatial data. That has meant separating tree crowns from depth maps over an orchard, tracing feature lines across a 3D-scanned heritage site, and training a transformer-based segmentation model to parse multimodal scans of horses for a research dataset.
I'm currently completing an MSc in Artificial Intelligence at the University of Cape Town and looking to continue this line of work at the PhD level — with particular interest in transformer-based 3D segmentation, multimodal registration, and applied computer vision for scientific and cultural-preservation use cases.
Developed backend and frontend components for web and mobile applications, contributing across the full product lifecycle from design to deployment.
Built and deployed the startup's main web application, focused on scalability and user-centered design.
Received mentorship on software engineering best practices and professional development guidance from industry engineers.
Enhanced orchard tree crown detection by integrating depth maps with Mask R-CNN, improving segmentation accuracy. Implemented in Python with OpenCV for agricultural canopy mapping.
Applied the Tensor Voting algorithm in Python for 3D heritage site analysis, supporting the Zamani Project's cultural preservation efforts.
View project →Built a virtual drone environment — from assembly to flight simulation — to reduce real-world training costs.
View on GitHub →Developed a church hymn app with 1,000+ users; built in Java, later migrated to Kotlin.
I'm looking for PhD positions in computer vision, 3D perception, and geometric deep learning starting Fall 2027. If my background is a fit for your lab, I'd welcome the conversation.