Arjun Haridas Pallath

 Design and implementation of 1D CNN model for classification of Human activity recognition from accelerometer data  Hyper parameter tuning of CNN model and comparison of model accuracies

 Design and implementation of 2D and 3D CNN models  Model training on neuroimaging dataset with more than 650 MRI scans  Generation of class activation and relevance maps; Saliency map, Guided Backpropagation, Grad-CAM and LRP  Hyper parameter tuning of CNN model and comparison of model accuracies and relevance maps  Interactive web-app as a guide for MRI relevance map visualizations using Bokeh framework

 Analysis of Genetic Algorithm used for maximizing capacity of wireless nodes to find regions of code that can be parallelized using gprof tool  Converted the serialized code to run in parallel on multi-threaded CPU cores using OpenMP to improve the execution runtime  Improved the runtime performance and achieved a speedup of 3.6 times speedup compared to the serial code execution time

 Configuration and maintaining a virtual cloud environment of 3000 servers using VMWare vSphere  Troubleshooting issues of virtual machines, Windows servers, and server hardware  Creation of Root Cause Analysis (RCA) and Standard Operating Procedure (SOP) documentations  Applying windows OS patching

Image processing, Natural Language processing, Embedded systems, Artificial Intelligence, Distributed Computing