Heartwatch has a proprietary product, a smart watch for multi-factor health monitoring and detection of anomalies. Signals captured by onboard sensors are transmitted to the server in real-time for analysis and machine learning-based prediction, and corresponding alerts are retransmitted to the app on the smartphone paired to the watch.

• Involved in signal analysis as well as design, implementation and optimization of corresponding machine learning applications for acceptable real-time response generation.
• Designing real-time test scenarios for testing the application.
• Extensive data analysis for determination of optimal time-series prediction algorithms.
• Unit testing as well as integration testing of various modules.
• Optimization of code for socket-based server-side communication.
• Member of Scrum team.

Developed multiple adaptive algorithms utilizing game-theoretic mechanisms to optimize bandwidth sharing and link formation in 4G and 5G wireless communication networks, with significantly low interference and high signal-to-noise ratio (SNR). Designed algorithms were tested on smartphones accessing a variety of multimedia applications with elastic bandwidth requirements, under fairly dynamic network scenarios (nodes randomly entering/exiting network).

• Involved in communication signal and data rate tracking and analysis in diverse network scenarios.
• Responsible for design, implementation and optimization of game-theoretic bandwidth sharing and link establishment mechanisms (through MATLAB, Java and C) to allow optimal network communication at low error rates.
• Conducting unit as well as integration tests.
• Conducting extensive simulations on MATLAB, Java and Python.
• Performing extensive data analysis for optimal mechanism identification, in Python.
• Real-world application testing.
• Writing research papers (three papers accepted, two in high-impact journals and one in a first-tier IEEE conference).
• Writing extensive research reports, documenting findings from analysis, preparing system design schemas and compiling results of real-world testing of applications.

Developed multiple low-complexity machine learning algorithms for prediction of meteorological phenomena, with special focus on rain events. Designed spline regression-based algorithms for predictions based on parameters such as exceedance and brightness temperature, under dynamic weather conditions

• Involved in communication signal and data rate tracking and analysis in dynamic weather conditions.
• Responsible for design, implementation and optimization of spline regression-based ML algorithm (through MATLAB and Java) to allow early prediction of meteorological events at low error rates.
• Conducting unit as well as integration tests for ML code.
• Conducting extensive simulations on MATLAB, Java and R.
• Performing extensive data analysis in R.
• Real-world testing of designed algorithms.
• Writing research papers (six papers accepted, three in reputed journals and three in first-tier IEEE/Springer conferences).
• Writing extensive research reports, documenting findings from analysis, preparing system design schemas and results of real-world testing of applications.
• Part of agile team.

Developed Bayesian Machine Learning algorithms for prediction of employee attrition for a retention management product.