• Designed and built 2 distributed machine learning frameworks, CarML and EdgeMLSim, to model machine learning using MXNet and simulate mobility of edge devices on real maps using SUMO• Researched extensions of the frameworks to: i) optimize placement of roadside units to increase performance by over 100%, ii) integrate vehicular clouds as edge servers to enable more efficient gradient collection from edge devices, iii) study real-time location-specific data collection, and iv) improve fault-tolerance• Implemented a Byzantine tolerant SGD algorithm, CRACAU, in an edge computing setting, proved its convergence, and evaluated its performance with MNIST and CIFAR-10 datasets• Evaluated Byzantine reliable broadcast algorithms for modern applications like cryptocurrency and Internet-of-Things using network topology built with Mininet

• Led weekly labs and discussions with over 20 students for Software Engineering, Algorithms, Computer Organization, and Computer Science I classes• Held weekly office hours to help students understand course materials, graded 25-50 submissions for each assignment, and created Python scripts to speed up grading of programming assignments