Agent Orchestrator

Job Title: Agent Orchestrator

Location: New  York City, NY (Hyrbid)
Duration: 12 Months
Work Hours: 40 hrs/week

Job Summary: The Agent Orchestrator architects and manages systems of multiple AI agents working in concert assigning tasks, managing inter-agent communication, handling failures, and ensuring the overall multi-agent pipeline delivers reliable, high-quality outcomes. As organizations move from single-agent deployments to agent ecosystems, this role becomes the conductor of the AI workforce.

Responsibilities:

• Design multi-agent architectures defining agent roles, task decomposition strategies, and coordination patterns (hierarchical, peer-to-peer, market-based)
• Implement orchestration layers that route tasks to the appropriate specialist agents based on capability, context, and load
• Build reliable inter-agent communication protocols including structured handoffs, context passing, and result validation
• Define and enforce agent boundaries preventing scope creep, unauthorized tool access, and cascading failures across agent networks
• Develop fallback and escalation logic ensuring human-in-the-loop checkpoints are triggered appropriately
• Monitor end-to-end pipeline performance across multi-agent workflows, identifying bottlenecks and failure points
• Govern agent-to-agent trust implementing authentication, permission scoping, and audit trails for agent actions
• Collaborate with Agent Engineers on individual agent capabilities; align with Product on workflow outcomes
• Maintain architecture documentation and decision records for all multi-agent systems
• Lead technical reviews for new agent integrations and evaluate third-party MCP servers for enterprise suitability

Required Qualifications:

• 5+ years in software or systems architecture; 2+ years in AI/agent systems
• Deep understanding of distributed systems, event-driven architectures, and workflow orchestration (Temporal, Airflow, or similar)
• Hands-on experience with multi-agent frameworks (LangGraph, CrewAI multi-agent, AutoGen group chat, or custom)
• Strong grasp of LLM behavior in agentic contexts — planning, tool use, reasoning, and failure modes
• Experience designing APIs and integration layers at enterprise scale
• Familiarity with MCP (Model Context Protocol) and agent-to-tool integration patterns
• Ability to design for reliability, observability, and auditability in non-deterministic systems
• Excellent technical communication skills — able to document complex architectures clearly

Preferred Skills:

• Experience with agent governance frameworks and enterprise AI policy
• Background in enterprise integration platforms (MuleSoft, Boomi, or similar)
• Published work or open-source contributions in multi-agent systems