AI算法工程师（Agent Harness系统优化）-即梦
Core
Optimize the Agent Harness system for the Dreamina creative agent to improve online execution efficiency, stability, and cost-effectiveness.
Role type
Senior IC LLM Agent Systems Engineer
Builds
Multi-model, multi-tool Agent systems for creative content generation
Domain
Generative AI / Large Language Models / Agent Systems
Deliverable
production ML models
Required skills
Python, Go, LLM inference optimization, context management, token efficiency, model routing, latency analysis, throughput optimization, cost analysis, observability, system stability
Preferred skills
C++ or Java, Prompt Cache strategies, long-context performance tuning, cross-team collaboration
Technologies
LLM Serving, Model Deployment, Prompt Cache, Token Mechanisms, Agent Frameworks
Responsibilities
Optimize multi-turn Agent call chains including LLM request orchestration and context reuse; Analyze performance metrics (TTFT, throughput, TPM, latency, cost) and establish optimization mechanisms; Explore dynamic model routing strategies based on task type and constraints; Build logging, tracing, and monitoring capabilities for rapid issue diagnosis; Collaborate with algorithm, model, and product teams to deploy optimizations.
Seniority
Senior, hands-on IC
