Welcome!

I develop deep learning models for 3D medical images at ImmersiveTouch. Previously, during my graduate studies at Rensselaer Polytechnic Institute, I worked on Natural Language Processing and Machine Learning. Before that, I graduated from Bangladesh University of Engineering and Technology with a B.Sc. in Electrical and Electronic Engineering.

In my current role, I train segmentation models for different anatomical structures in CT scans. These segmentations are used to create 3D models of relevant organs for surgical planning. I have also developed landmark detection models for head CT scans. Before this, I did research on topics in representation learning, efficient models and information extraction. I have worked on domain adaptation of neural models for causal information extraction. I have also worked on multi-sense embeddings to address the meaning conflation problem in word embeddings. In this project, I implemented a knowledge distillation method to transfer the contextual information from pre-trained language models into multi-sense embeddings. During my internship at IBM Research, I worked on information extraction from structured documents such as scanned images, pdfs etc.