AI Solution Architect – Agentic AI

About the role

We are seeking an AI Solution Architect specialized in agentic and multi‑agent systems to design and govern the orchestration layer, memory architecture, and tool‑use permissioning that keep autonomous AI systems reliable and safe.

Key responsibilities

• Design multi‑agent architectures, defining agent roles, coordination patterns, and handoff logic using frameworks such as LangGraph, AutoGen, and CrewAI.
• Own the end‑to‑end agentic workflow design, including task decomposition, tool‑calling, and state management with knowledge‑graph‑based reasoning.
• Define and implement guardrails, kill switches, and escalation paths for agent failures.
• Architect evaluation frameworks in partnership with AI Assurance/QA teams.
• Lead technical design reviews and ensure fallback and human‑in‑the‑loop patterns for high‑stakes decisions.
• Mentor GenAI engineers on architecture patterns and review their designs.
• Present risk/benefit trade‑offs to technical and business stakeholders.
• Govern architecture standards across the practice and facilitate cross‑team reviews with Security, Governance, and Platform Engineering.

Required profile

• 10–15+ years of software/AI architecture experience, with at least 2 years focused on agentic or multi‑agent systems.
• Deep knowledge of agent orchestration frameworks, failure modes, and managed agentic platforms (Azure AI Foundry, AWS Bedrock Agents, Google Vertex AI Agent Builder).
• Strong systems‑design fundamentals in distributed systems, state machines, and event‑driven architecture.
• Experience designing memory architectures (vector and graph‑based) for long‑running agents.
• Proven ability to defend architectural decisions under adversarial technical challenge.
• Track record of mentoring technical teams and owning production‑grade agentic solutions.

Required skills

• LangGraph, AutoGen, CrewAI
• Microsoft Azure AI Foundry Agent Service
• AWS Bedrock Agents / Strands Agents
• Google Vertex AI Agent Builder
• Vector databases and graph databases for memory management
• Distributed systems, state machines, event‑driven architecture
• Agent evaluation methodologies and benchmarking
• Guardrail and governance tooling for agentic systems