CV

Independent Researcher

Independent

• Developing an explainable AI method for video classification models based on sufficient input subsets, isolating the minimal input frames that drive a model’s decision (manuscript under review, WACV 2027).

Research Scholar

University of Iowa

• Labelled a dataset and built a spatio-temporal multimodal deep learning framework for echocardiogram classification, integrating clinical data with video sequences to improve detection of Takotsubo cardiomyopathy.
• Scheduled GPU-heavy training on an SGE-based high-performance cluster using CUDA and PyTorch.

PhD Researcher

University of Virginia

• Developed a novel single-image depth estimation method exploiting defocus blur, generalizing across camera settings and outperforming prior state of the art on cross-camera evaluation.
• Designed CNN-based techniques for noise-robust audio emotion recognition, validated across multiple public benchmark datasets.
• Formulated a few-shot learning method for wearable-sensor human activity recognition, outperforming existing approaches on three public datasets.
• Co-designed and deployed an in-home care system for Alzheimer’s patients and caregivers; built the AWS-based real-time data pipeline and a Slack-based alerting system for the care team.
• Led design and data collection for a multimodal (video, IMU, audio) CPR performance dataset across 23 participants.
• Trained models on a Slurm-based HPC cluster with CUDA, PyTorch, Keras, OpenCV and MATLAB.

Machine Learning Intern

Elder Research

• Explored neural differential equation solving in Julia for modeling space object dynamics; programmed C libraries to extract training data from NASA’s SPICE toolkit.
• Developed and validated a neural differential equation solver for space object positioning and communicated the results to the team.

Data Science Engineer

CodeGen Sri Lanka

• Built a two-tiered image classification system for customer preference analysis, a Naïve Bayes filter with over 90% accuracy on cartoon images followed by a CNN (Caffe).
• Developed customer demand prediction models from historical and weather data, comparing LSTM, random forest and XGBoost and selecting LSTM.
• Prototyped an ALS-based recommender for personalized suggestions.

Research-based MS Student

Imperial College London

• Built Naïve Bayes classifiers for real-time hand motion classification from smartwatch data, with companion Android apps; led a seven-participant study on tooth brushing and hand washing.

PhD

University of Virginia

Computer Science

MS

Imperial College London

Medical Robotics and Image Guided Intervention

• Faculty of Medicine Dean’s Prize, awarded to the top student of the program.

BSc

University of Moratuwa

Mechanical Engineering

Camera-Independent Single Image Depth Estimation from Defocus Blur

WACV 2024 · Wijayasingha L., Alemzadeh H., Stankovic J.

Generalized Few-Shot Learning for Wearable Sensor-based Human Activity Recognition

IEEE PerCom Workshops 2022 · Wijayasingha L., Stankovic J.

Robustness to Noise for Speech Emotion Classification using CNNs and Attention Mechanisms

Smart Health 2021 · Wijayasingha L., Stankovic J.

A Wearable Sensing Framework for Improving Personal and Oral Hygiene for People with Developmental Disabilities

IEEE Wireless Health 2016 · Wijayasingha L., Lo B.

EgoEMS: A High-Fidelity Multimodal Egocentric Dataset for Cognitive Assistance in Emergency Medical Services

AAAI 2026 · co-author

EgoCap and EgoFormer: First-person Image Captioning with Context Fusion

Pattern Recognition Letters 2024 · co-author

University of Virginia Computer Science Fellowship

University of Virginia

Faculty of Medicine Dean's Prize

Imperial College London

Awarded to the top student of the program.