Harness/Runtime架构师（小花项目） - 抖音研发
Core
Design and optimize the underlying architecture for the Xiaohua Agent Core Loop execution engine and Agent Runtime, including Reason-Act orchestration, context management, and capability routing.
Role type
Senior IC system architect (AI Agent Runtime)
Builds
High-scale AI agent execution engines and runtime infrastructure for Douyin's products
Domain
AI Agents, Large Language Models, Search & Recommendation
Deliverable
production ML models
Required skills
Large model application architecture, AI Agent paradigms (ReAct, Function Calling, RAG, Multi-Agent), System architecture design, Cost-latency-effect trade-off optimization, Multimodal technology integration
Preferred skills
Search/Recommendation/Ad systems experience, Multimodal experience, 0-to-1 complex system building experience
Technologies
ReAct, Function Calling, RAG, Multi-Agent, Multimodal LLMs
Responsibilities
Design and optimize the Agent Core Loop execution engine architecture; Design Agent Runtime core mechanisms including Reason-Act orchestration and context management; Optimize agent execution efficiency balancing cost, latency, and performance; Integrate multimodal technology systems; Track and evolve industry Agent Runtime technologies
Seniority
Senior, hands-on IC
