Updated compute infrastructure to Ubuntu 18.04 while maintaining functionality of tools and pipelines for medical image processing and machine learning, transitioning all 12 workstations to LTS and establishing 2 new compute server nodes.

Conducted instruction through office hours, labs, discussion sessions, online assistance, and grading feedback, allowing students to access and grasp CS 107/107E Computer Systems material easily and efficiently: C, ASM (ARM/x86), systems, sensors, graphics, coding style, etc.

Developed solutions to further optimize trajectory prediction neural network for FP16 and INT8 inference on Drive AGX platform, resulting in over 2.0x speedup over previously optimized models.

Created deep learning workflow from scratch for screening GPUs by capturing frames from traces with DirectX, training networks, and setting performance metrics and benchmarks (Python and C++).