The radio signals received on Earth are contaminated by many man-made radio sources.
These radio interference sources range from a spark plug of a bike to an AC powerline.
This project aims to develop a radio interferometer pipeline which will simulate the
astronomical radio signals and its interference using functional data analysis. The
interference noise would be then removed using statistical tools and techniques.

The rocks found in the Deccan Plateau region can be classified into 4 main
categories. These 4 main categories can be further divided into 19 sub-classes.
This project aims to build a GPU based CNN model to classify the rocks of Deccan
Plateau to a fair accuracy with a limited set of images of the rocks.

Evolutionary Algorithms are one of the powerful metaheuristic techniques used for
optimization. The aim of the project was to generate a target image using only
polygons. The complete image data was encoded as a binary sequence and
customized techniques of crossover and mutation were applied for reconstruction.

Predicting depth from the color images using custom made loss function.

I was involved in the end to end product development for activity recognition in the logistics, airport and homeland security industry. I worked in ideation, design, development and deployment of the software on scale.