EN — AI engineer who turns large language models into autonomous agents that can plan, call tools, run code in sandboxes, search knowledge bases, and remember across sessions. By day I ship a production LLM-agent platform; by night I dig into agent-harness research.
中文 — AI 工程师，专注把大语言模型变成会自己规划、调工具、跑沙盒代码、查知识库、跨会话记忆的智能体。白天打磨生产级 LLM Agent 平台，晚上研究 Agent Harness 相关 Research。
class LuhaoZhu:
def __init__(self):
self.role = "LLM Agent Platform Engineer"
self.company = "CCAI"
self.location = "Ningbo, China 🇨🇳"
self.building = ["ReAct agents", "MCP tool ecosystem", "sandboxed code-exec"]
self.researching = ["agent harness design", "tool-use & planning", "eval & reliability"]
self.motto = "Make the model do the work — reliably, at scale."
def current_focus(self) -> list[str]:
return [
"🔁 Migrating an agent runtime onto AgentScope 2.0",
"🧰 Growing an MCP tool fleet (web search, office docs, sandboxes)",
"🗂️ RAG knowledge bases + cross-session memory (mem0)",
"🏗️ Carving a 100k-line monolith into clean layered packages",
]Languages
LLM / Agents
Backend / Infra
AgentScope 2.0 · MCP servers · sandboxed code execution · RAG + mem0 memory · multimodal LLMs
"The best agent is the one you forget is an agent — it just gets the job done." "最好的智能体，是你忘了它是智能体——它只是把事办成了。"
⭐️ From Luhaozhu