Rate My Professor Kevin Dsouza

Kevin Bradley Dsouza is a Postdoctoral Fellow at the University of Waterloo. He earned a Ph.D. in Electrical and Computer Engineering from the University of British Columbia in 2023, an M.A.Sc. in Electrical and Computer Engineering from the University of British Columbia in 2018, and a B.Tech. in Electronics and Communication Engineering from the National Institute of Technology Karnataka in 2017. His work centers on artificial intelligence for science, machine learning, and computational sciences. Dsouza previously served as a Machine Learning Engineer at Gandeeva Therapeutics, where he applied sequence and structure-based machine learning, energy metrics, and Bayesian optimization to antibody design and protein affinity maturation. He completed his doctoral research on representation learning strategies for the epigenome and chromatin structure using recurrent neural models at the Libbrecht lab. Dsouza holds an NSERC Postdoctoral Fellowship and has consulted on multiple projects in AI for science during his time at the University of Waterloo. His publications include the 2022 paper "Learning representations of chromatin contacts using a recurrent neural network identifies genomic drivers of conformation" in Nature Communications, the 2021 paper "Latent representation of the human pan-celltype epigenome through a deep recurrent neural network" in IEEE/ACM Transactions on Computational Biology and Bioinformatics, and the 2025 paper "Assessing the climate benefits of afforestation in the Canadian Northern Boreal and Southern Arctic" in Nature Communications. Additional works address climate-adaptive boreal forest management and afforestation benefits. Dsouza has delivered talks on representation learning for biology at conferences including 4DN, RECOMB, and MLCB. He is the founder of Symbolia Labs, which develops AI-native platforms for science, health, climate, and enterprise applications.