Optimized the GPU power estimator by reducing the simulation runtime by 46.5% and introducing precise error handling for efficient debugging

Developed loop detection of kernel executions for easier workload analysis

Implemented hooks for NVSwitch and NVLink power model for system power estimator's future integration

Led the development and deployment of Point-of-Interest (POI) tagging recommendation services, which serves over 7 billion creator videos

Shared knowledge and experience within the team by documenting technology stacks and solutions to issues

Created our own dataset with 150 videos containing 27 different types of activities using Wyze cam and open source data

Winner of 2022 ENGINE Capstone program, ranked 3rd out of 48 teams