Graduate Student

• Sep 2022 - Present

Graduate Teaching Associate

• Jan 2023 - May 2023

Conducted classes and covered lecture materials for a Basic Algorithms course for undergraduate students. Helped design assignments and tests and conducted office hours to help students better understand the lecture material

Research Intern

• May 2021 - Aug 2021

Designed novel federated learning techniques for application in brain tumor segmentation. Parallelized the training of federated models on the Compute Canada cluster for a threefold speedup. Achieved a Dice Similarity Coefficient of 0.674 for a federated model by utilizing novel aggregation functions and variable local training which is comparable to the performance of a central model trained on pooled data. Designed novel federated learning techniques for application in brain tumor segmentation. Parallelized the training of federated models on the Compute Canada cluster for a threefold speedup. Achieved a Dice Similarity Coefficient of 0.674 for a federated model by utilizing novel aggregation functions and variable local training which is comparable to the performance of a central model trained on pooled data.

Software Development Intern

• May 2020 - Jul 2020

Implemented Kernel Prediction Networks, an Auto-Encoder based CNN architecture in Tensorflow and tested its performance. Compared performance of four channel Bayer (Raw) and three-channel RGB images for Denoising. Experimented with combinations of Perceptual, L1 and Gradient Loss for Video Denoising and compared their performance. Investigated the impact of additional noise estimates on Video Denoising results and their impact on Model Performance. Implemented Kernel Prediction Networks, an Auto-Encoder based CNN architecture in Tensorflow and tested its performance. Compared performance of four channel Bayer (Raw) and three-channel RGB images for Denoising. Experimented with combinations of Perceptual, L1 and Gradient Loss for Video Denoising and compared their performance. Investigated the impact of additional noise estimates on Video Denoising results and their impact on Model Performance.