With the advent of LLMs, knowledge of system infrastructure, especially on parallel and distributed computing, is becoming increasingly important for NLP research. In-depth understanding of distributed training frameworks enables researchers to predict the costs of their training and perform tailored optimization to improve latency and memory efficiency. In my opinion, this course plays a crucial role in building the foundation of system education for future NLP researchers. Therefore, throughout the preparation of lecture materials, lab assignments and office hour recitation, I have not only been polishing the delivery of traditional system knowledge, but also preparing my students for the fascinating world of LLMs and machine learning systems. In my anticipated academic career, I am also eager and well-equipped to instruct future machine learning researchers on both data-centric topics and beyond.