豆包大模型Agent算法工程师 - 火山方舟
Core
Designing and optimizing the architecture and runtime mechanisms for managed AI agents and harnesses to enable enterprise AI application deployment.
Role type
Senior IC Agent Algorithm Engineer (Harness & Multi-Agent Systems)
Builds
Managed-Agent platform, Harness runtime, multi-agent collaboration schedulers, and intelligent auxiliary capabilities.
Domain
Large Language Models (LLM) and AI Agent Systems
Deliverable
production ML models
Required skills
Agent Harness optimization, Multi-Agent collaboration scheduling, LLM API integration, ToolUse and FunctionCalling, RAG retrieval, Agent evaluation methodologies, ReAct patterns, KV Cache management, MCP and A2A protocols, Model selection and Advisor logic, Code evolution techniques (AutoPE, Memory Evolve).
Preferred skills
Experience with OpenClaw, HermesAgent, Codex, ClaudeCode, AlphaEvolve, Hyper-Agents, Ralph loop implementation.
Responsibilities
Selecting and designing the technical architecture for the Harness system; Optimizing Harness runtime modes, built-in tools, and context mechanisms; Building end-to-end evaluation frameworks for Harness iterations; Collaborating with model teams to achieve co-evolution of Seed Models and Harnesses.
Seniority
Senior, hands-on IC