-Tech lead for project implementing new component of Microsoft Teams-Coordinated a large, cross-team effort to resolve latency issues of performance critical machine learning inferencing component of pipeline to enable release of product to production-Rearchitected interface between 2 machine learning inferencing platforms to enable scaling of an email understanding service in Outlook-Improved monitoring of email understanding system to facilitate identification of performance issues of an email understanding platform

-designed and developed a REST API framework using node.js to facilitate automated software testing of set top box playback related functionalities.-developed software automation solutions in Javascript and LabView to automate the process of testing set top box signal modulators.

-Software Engineer II-Implemented algorithm in C++ to detect possible targets from hyperspectral sensors placed on military aircraft. Algorithm detects hyperspectral image data pixels that are spectrally different from surrounding pixels and grows the regions of interest around these pixels classified as highly likely to correspond to targets. Optimized algorithm by utilizing more efficient data structures to store pixel data. Optimizations increased algorithm’s efficiency by 50%.-Optimized and resolved memory leak issues of proprietary hyperspectral data processing algorithm. Devised an efficient method to perform multiplications between large matrices. Optimizations increased efficiency of matrix multiplication by a factor of 10.-Lead Software Developer – Atmospheric Correction Software Project-Optimized and modularized a large-scale hyperspectral data processing software containing over 50,000 lines of C++ source code to make it more efficient, modular, and flexible.-Designed and developed an equation parsing algorithm to enable the end user to input arbitrary metrics for identifying target pixels-Built testing framework to automate process of testing newly developed software on a large number of datasets.Log File Parser Project-Developed a log file search tool for hyperspectral data analysts to quickly extract relevant information out of error logs-Software tool was developed in C++ using the QT framework

-Developed C++ multithreaded testing application on 3M Cogent’s fingerprint sensing devices to detect possible conflicts between fingerprint sensing and WIFI data transmission processes of the operating system.-Developed C software interfacing with an iris recognition camera to support development of iris recognition algorithms.-Provided technical solutions to customers regarding issues with the fingerprint sensing mobile devices.

-Developed part of Raven, a proprietary signal processing engine.-Programmed in C the multiresolution analysis, subband filtering, and synthesis functionalities of the Raven engine.-Tested company’s room correction equalization algorithm before delivery to customers

-Modified an external driver device written in C for wireless cards with Atheros chipsets to gather clear channel statistics from a certain computer(node) in a network. Statistics gathered from each node in a network are sent to other nodes, enabling the creation of efficient algorithms utilizing the statistics to increase throughput in the network.-Designed and implemented MATLAB interpolation algorithm with least squared error techniques to estimate continuous time signal, which enabled estimation of packet collision probabilities and development of throughput enhancement techniques.-Research team members utilized work in this project in a paper submitted to the 2011 IEEE Wireless Communications and Networking Conference (Link: http://www-video.eecs.berkeley.edu/papers/mkrishnan/mwc-wcnc11.pdf)

-Created a feedback-based distributed beamforming algorithm that maximizes the power of a received signal by deterministically matching the phases of signals transmitted over a random channel by distributed transmitters.-Algorithm increased energy efficiency of signal transmitters.-Wrote a technical report on research titled “Performance Analysis of a 1-bit Feedback Beamforming Algorithm” (Link: http://www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-161.pdf)