Zedong Peng

Research: I am interested in the development of large language models, their systems optimization and applications.

Recently, I was inspired by Karpathy’s LLM Wiki and build my own [Research Wiki]. Bye Zotero :). Karpathy also mentioned parametric memory, a concept that I attempted but failed [blog post]. A naive system idea [Blog: linear attn bottleneck on Blackwell GPU].

Side interests: [llama.v] LLM inference on FPGA. [benchmark-grinder] a skill let your Claude Code autonomously improve your project performance.

Previously: I was a research intern at the NCTI for EDA (2025), where I worked with Prof. Qiang Xu on LLM4EDA’s HLS part [blog post]. I was a research intern at Intel (2023-2024), lead by Yuan Zhou on [Gluten] and [vLLM (CPU)]. I obtained my Bachelor’s degree (with honors) in Electronic Engineering from Shanghai Jiao Tong University (2024), where I worked with Prof. Liyao Xiang.

I consider myself a self-driven and curious researcher, seeking 2026 summer research internship opportunities!