Designing and implementing efficient training and evaluation pipelines for deep learning models in video computer vision.

Implemented scheduling algorithms in a NIC that can deliver total packet order at 100 Gbps, outperforming state-of-the-art designs in throughput. Used DPDK libraries and drivers for fast packet processing in high-speed networking applications.

Designed and implemented in Python an access planning tool for optimally deploying the optical front, mid, and backhaul of 5G networks. Modeled residential and 5G mobile users, base station types, wireless environments, and functional split choices.

Implemented a state-of-the-art intent-based allocation algorithm for cloud computing resources using Q-Learning. Implemented and evaluated MinIO, IPFS, and BigchainDB for edge computing storage in IoT environments.

Implemented an image segmentation model based on attention recurrent residual U-Net for visual road crack inspection.