- Automotive Compute

- Analyzed and profiled various state-of-the-art DL algorithms for Perception (CNNs, ViTs, Efficient Transformers) and Prediction (CNNs, LSTMs, Transformers, GNNs) stacks of Autonomous Driving Pipeline by simulating on cycle-accurate simulators to run the above DL models and extended open-source DNN inference accelerator simulation tools such as STONNE, MAESTRO.- Compared performances of the above DL models (Inference cycles, PE utilization) on Systolic array-like (TPU), Parallel vector-like (NVDLA, MAERI), Marvell ML Accelerator architectures with different dataflows.- Built a DNN simulation pipeline for Perception, Prediction stack to facilitate the benchmarking process of new DL models for Automotive to match DL compute in Tera Ops Per Sec (TOPS) given the limited Sq. mm area of the ML IP.

- Assisted students with programming exercises and concepts in weekly discussion sessions and office hours- Contributed to course development by improving the course material such as MPs

- Researched public safety using Computer Vision to detect human movement in human-robot collaborative workspaces.- Created IoT sensor monitoring web application that stores and manages different sensor data in real time.

- Developed a web app that enables students to simulate IoT circuits and submit their circuits for grading. Used React.js on frontend and built RESTful API using Flask with MongoDB and deployed on AWS.

- Designed Core Router (CR), Edge Router (ER), Access Router (AR) networks to prevent network failures through multiple backup connections between routers.- Managed static/bgp routing tables for WAN/LAN and access control lists (ACL) on routers and switches.- Consulted clients by providing network solutions based on the clients' traffic usages.- Maintained and troubleshooted the Military Broadband Convergence Network covering Korea, UAE, Lebanon.