Jul 2017 – Sep 2017 · 2 mos · Berlin Area, Germany

• Implemented parallelized algorithms for distribution samplers with GPU support to reduce operational latency
• Used cuRAND library (CUDA Toolkit) to develop different samplers on C++ with a Python testing environment
• Assumed full responsibility of the project and contributed to MXNet, an open-source library for Deep Learning
• Also developed a box-cox transformation operator using MXNet, to be used by other teams to normalize samples

Jun 2015 – Aug 2016 · 1 yr 2 mos · Mumbai Area, India

• Analyzed consumer behavior and made prediction models based on it; targeted consumers more likely to comply
• Used regression analysis and other statistical models to increase marketing efficacy and induce desired response
• Built a model to increase consumer transaction frequency and repeat purchase behavior for leading retail merchants to generate a revenue of approximately INR 5 million in the period April-June 2016

May 2014 – Jul 2014 · 2 mos · Bengaluru Area, India

• Enabled use of 2D accelerator design for UIs with same performance as 3D Graphics engine but at a lower power
• Modeled & implemented an innovative blending technique to address issues like screen freezing in smartphones
• Abstracted & verified RTLs in SystemVerilog for the above technique; ensured that it satisfies industry standards