I'm Kuo Li, an undergraduate at Tongji University, working on full-stack machine learning systems.
My interests include:
- LLM training, inference & serving — KV cache optimization and long-context inference.
- Kernels & compilers — LLM-driven kernel generation, DL compilers, CUDA, and MegaKernels.
- Efficient AI — sparse attention and hardware–software co-design.
- AI & agent infrastructure — agent runtimes and operating systems.
I'm also exploring embodied AI, generative models, and AI for software engineering.
Building systems that make AI faster, cheaper, and more reliable.
CUDA · vLLM · Transformers · RISC-V
- 🥇 National First Prize — Global Campus AI Algorithm Challenge
- 🥈 International Silver Medal — iGEM