I write C++ and CUDA, with an ACM silver medal background.

Right now I am working on ContextOS and learning LLM inference from the bottom up: GPU kernels, transformer internals, generation loops, and KV cache.

• GPT2.cu
  Handwritten C++/CUDA GPT-2 inference with real Hugging Face weights, greedy generation, prefill/decode, and KV cache.
• ContextOS
  Experimental personal context system for project memory, technical notes, and agent workflows.

• 从零手写 CUDA GPT-2 推理：从 full recompute 到 KV cache

• ACM silver medal
• C++ algorithms and data structures
• CUDA and GPU programming
• LLM inference systems

• Homepage: SyaOtiLan.github.io
• GitHub: github.com/SyaOtiLan