Sep 2017 – May 2018 · 8 mos · Greater Madison Area

• Architectural choices in deep learning
• Hyperparameter selection using input data statistics, convergence goodness and expected gradient, within given resource constraints
• Aimed at tackling the issue of hand tuning neural network hyperparameters through trial and error, specifically for brain images and standard image datasets (MNIST, Cifar10, SVHN, MRBI)

Jul 2015 – Jul 2017 · 2 yrs · Bengaluru, Karnataka, India

• As part of the Ad Selection team in Bing Ads, built algorithms based on machine learning and natural language processing to select relevant Ads (Both Product and Text Ads) for search queries on Bing.

Jul 2014 – Jun 2015 · 11 mos · Bengaluru, Karnataka, India

• As part of the Embedded Multimedia for Automotive team, built hardware accelerated multimedia tools using GStreamer

Jun 2013 – Jul 2013 · 1 mo · Greater Hyderabad Area

• Implemented Dynamic Time Warping with Affinity Propagation algorithm for recognizing gestures using accelerometer readings from haptic feedback insole (developed by the company) with an accuracy of ~80%, as a Java and an Android application