💼 Experience

Postdoctoral Researcher · full time

Royal Melbourne Institute of Technology (RMIT), Melbourne, Australia.

• Developed and deployed an AI-driven digital twin integrated with agentic AI frameworks to support manufacturing production monitoring, anomaly detection, operational decision-making, and business process optimisation at Kilcoy Global Foods, Australia.
• Working on a medical video analysis project to analysis surgical videos and provide real-time feedback to surgeons using computer vision and deep learning techniques.

Research Fellow · full time

Advanced Multimedia Research Lab, ICT Research Institute, University of Wollongong, Australia

Designed and deployed a real-time computer vision system integrating Jetson Orin AGX, Intel RealSense L515, and GPT-4-assisted workflows for upper-body motion analysis and automated clinical reporting.

PhD Candidate, Researcher · full time

Advanced Multimedia Research Lab, ICT Research Institute, University of Wollongong, Australia

• Designed and implemented end-to-end human skeleton motion analysis models using Graph Convolutional Networks, Transformers, Masked Autoencoders, diffusion models, supervised learning, and self-supervised learning.
• Developed scalable Linux-based HPC pipelines for large-scale benchmarking, reproducible experimentation, model validation, ablation studies, and the complete research lifecycle from problem formulation and data preparation to evaluation and publication.
• Published research in Q1 journals, presented findings at conferences, collaborated across interdisciplinary research teams,

Research Engineer · part time

Department of Electrical and Electronic Engineering, University of Peradeniya, Sri Lanka

• Developed a grant-supported edge-AI system for night-time wild elephant detection on railway tracks using YOLO, customised PyTorch models, Jetson AGX Xavier, ROS, and real-time RGB video streaming from Reolink cameras using C++ and Python.
• Designed an end-to-end multitask pipeline for video acquisition, low-light inference, decision-making, alert transmission, and mobile visualisation; also contributed to NVIDIA grant proposal development, with the project receiving additional IEEE RAS SIGHT funding.
• Developed MRI brain tumour localisation, classification, and segmentation pipelines using CNNs, Faster R-CNN, and Chan–Vese active-contour refinement, achieving a Dice score of 0.92.

Temporary Instructor · full time

Department of Electrical and Electronic Engineering, University of Peradeniya, Sri Lanka

Associate Electronic Engineer

Zone24x7 (Pvt) Ltd, Colombo, Sri Lanka

Worked on the AZIRO Robotics project.

Engineer Trainee

Industrial Control System, Colombo, Sri Lanka

Universal Robots UR arm control; a vision-based error tracking and packing system for Phoenix Plastics.

Instructor · part-time teaching assistant

Department of Electrical and Electronic Engineering, University of Peradeniya, Sri Lanka

Engineer Trainee

Dialog Axiata PLC, Colombo, Sri Lanka

Transmission network planning; capacity planning for 2G, 3G and 4G; optical fibre communication and GPON technology testing.