- 🔭 Working on LLM inference & AI infrastructure at
EVAS Intelligence Co.,Ltd - 🧠 Deep into compilers (LLVM / RISC-V), AI Operators, KV Cache, MoE architectures & Agent systems
- 📝 I write about compilers and LLM internals on my blog → z3475.work
- 💬 Ask me about C++, LLVM, CUDA operators, LLM serving
- 🏆 ICPC competitor, ACM problem setter mindset
- ⚡ Forward to AGI.
| Project | What it does |
|---|---|
| 🕸️ pcl | Prompt Control Language — a template DSL for orchestrating LLM agents |
| 🏔️ gitmount | Browse every git branch, tag and commit as a read-only filesystem (C++) |
| 🐍 MCP-WASMPython-Runner | A safe sandboxed MCP Python runner with Docker |
| ⚡ libzio | Coroutine library built on C++20 coroutines + io_uring |
| 🌐 HNUSTnet | Campus network auto-connect tool (C++) |
| 📚 ATL | Algorithm template library for ACM/ICPC |
- 📝 LLVM RISC-V 后端：从 IR 到 MIR 再到汇编 — 2026-09-13
- 📝 LLVM IR 中端 Pass 的设计目标与运行原理 — 2026-09-12
- 📝 Agent原理介绍 — 2026-09-02
- 📝 KV Cache 笔记 — 2026-08-30
- 📝 从llama2到Qwen3.5-MoE，再到DeepSeek-V4-Flash——架构变迁笔记 — 2026-08-29
↻ Auto-refreshed every 6 hours by blog-post-workflow · 📜 RSS
Languages
Compilers & AI Infra
Toolchain & Platforms
⭐ From OrbitZore with 💙 · Last refreshed by GitHub Actions