Solutions Architect

Agentic AI Solution Architect (Lead)- London

Role

Purpose Own the architecture and delivery of enterprise-grade multi-agent systems on a hybrid AWS and Open Source stack, ensuring scalability, security, and production readiness. Key Responsibilities Architecture and System Design Define multi-agent system architecture (planner, executor, reviewer, memory) Author Architecture Decision Records (ADRs) for: AWS Bedrock vs open-source models Hosted vs self-managed inference Design event-driven orchestration using Step Functions and/or LangGraph Hybrid AI Strategy Implement model routing strategies: High-complexity workflows to managed LLM platforms (e.g., Bedrock) Cost-sensitive tasks to open-source models (e.g., Llama, Mistral) Define fallback and failover logic LLMOps and Platform Thinking Architect: Model and prompt versioning pipelines Evaluation pipelines Red-teaming frameworks Define observability metrics such as latency, token cost, and hallucination rates Security and Governance Design IAM-based agent permissions and secure API/tool access Ensure compliance with enterprise security and governance standards Stakeholder Management Translate business problems into technical architecture Align with security, infrastructure, and business teams Technical Stack Must Have Strong AWS experience (at least 3–4 of the following): Bedrock (or equivalent), Lambda, API Gateway, Step Functions, IAM, VPC Experience with at least one agent framework: LangChain or LangGraph Infrastructure as Code: Terraform or AWS CDK Strong understanding of distributed systems, event-driven architecture, and RAG patterns Good to Have CrewAI or AutoGen Observability tools such as LangSmith or equivalent Knowledge graphs or GraphRAG exposure Exposure to GPU or NVIDIA ecosystem Experience with hosting open-source models