Developed an AI inferencing server using C++/STL leveraging gRPC, and asynchronous callbacks to facilitate client-servercommunication across multiple chipsets from Intel, and Qualcomm (OpenVINO and Snapdragon Neural Processing Engine).• Contributed to the async multi-threaded Python Machine Learning Pipeline Execution Engine, improved code quality, and followedagile practices.• Responsible for the development and implementation of REST API, wrote unit and integration tests using pytest, and bug fixes.• Worked as part of porting the Execution Engine to rust using tokio-rs.• Implemented a safety fallback feature for the Execution Engine using a time-based leaky queue, thus improving system reliability tomaintain stability during high inflow of images.• Designed and integrated Megh Model Server into Nimble Execution Engine, which is inspired by NVIDIA Triton Inference Server.• Contributed to many bug fixes and development of Golang based CLI installer.

- Designed problems, and resolved student doubts for CSC-505, a graduate-level course on algorithm design covering concepts likeGraph Path Finding, NP-Completeness and approximation algorithms.

- Developed an audio conferencing application with ReactJS and Firebase, featuring multiple public and private spaces, and receiveda 100,000 INR funding grant from the Government of India during the COVID-19 pandemic to aid communication in war rooms- Led the end-to-end development of the application, taking it from scratch to beta deployment, and successfully implemented it forefficient communication in COVID-19 war rooms, supporting volunteers in their fight against the disease.