Designed and built various GPU based modules for Adaptive Real Time Machine Learning (ARTML) project.Compared the speeds of various GPU libraries such as TensorFlow, Numba (CUDA kernels), CuPy, PyTorch and Gnumpy for ARTML GPU implementation.Identified CuPy as the fastest library which gave 50% speed improvement with respect to the CPU variant.Vectorized CPU computations for Basic Element Table module and achieved a 90% enhancement in speed as compared to the original code.

Acquired training in technologies such as TypeScript, AngularJS, NodeJS and ExpressJS.Headed a team of four in implementing the knowledge of these technologies to develop web applications.