Sachith Seneviratne

Dr. Sachith Seneviratne is a Research Fellow at the University of Melbourne's Melbourne School of Design, specializing in Computer Vision and Health. His research bridges Machine Learning, Urban Design, and Public Health, with a focus on Deep Learning, Contrastive Representation Learning, and Agent-Based Modelling. He holds a PhD in Machine Learning from Monash University, exploring automated parallelized machine learning models. His work has collaborations with NASA, Harvard University, and Currnt Research Hub. He is affiliated with the Transport, Health and Urban Design (THUD) research group, focusing on 'Data and Value,' 'Future Cities,' and 'Healthy Communities and Infrastructure.' His recent publications span facial deepfake detection, urban design impacts on health, and automated architecture generation using LLMs.

Research interests include generative design, automated infrastructure planning, and the ethical implications of facial recognition under privacy constraints (e.g., masked faces). He has pioneered methods for collecting population-level bicycling data via smartphones/Bluetooth beacons and developed novel architectures like Arch-LLM for neural architecture generation. His work on subgrid-informed flood mapping and physics-informed neural networks demonstrates cross-disciplinary innovation.

Consulting work includes machine learning projects for NASA and Harvard. His lab teams focus on creating interpretable AI systems for environmental and health applications, such as Ginn-kan for physics simulations and Ginn-lp for symbolic equation discovery. Current projects aim to quantify urban design's impact on non-communicable diseases and improve walkability through micro-scaled spatio-temporal analysis.