Projects
Projects
Selected research and engineering work.
PriniaXR — Differentially Private Face Recognition
- Built a differentially private face recognition pipeline integrated into XR environments, balancing privacy with identification performance.
- Designed end-to-end Python workflows for data preprocessing, model training, evaluation and inference, emphasising privacy-preserving transforms and secure handling of sensitive data.
PyTorch · scikit-learn · Python
Poultryfi — Egg Counting & Grading
- Developed computer vision models that automated egg counting and quality grading with better-than-human accuracy, reducing manual effort and cost.
- Built Python batch pipelines to ingest imagery and generate structured production records for downstream reporting; deployed to several small-scale farms.
- Compressed large object detection models using knowledge distillation and quantisation for optimal performance on edge devices.
PyTorch · TensorFlow · Python · Edge deployment
3D-Motaas — Motion Capture Search Engine
- Implemented autoencoder-based motion similarity and feature engineering to enable fast retrieval across motion-capture datasets.
- Created Python tooling to clean, transform and index sequences for repeatable offline/batch inference and querying.
- Built a horizontally scalable distributed pipeline to parallelise motion processing across machines.
PyTorch · Celery · Docker · Python
Publications
EurIPS 2025 Workshop: Epistemic Intelligence in Machine Learning
International Conference on Computer and Applications (ICCA), Cairo, Egypt · pp. 1–6
Real-Time Egg Detection Using Edge Computer Vision
1st International Conference on Accessible Digital Agriculture Technologies (CADAT), Valencia, Spain · pp. 16–22