Agentic AI & ERP Integration - AI Architect

Role DescriptionWe are seeking a senior AI Architect to design and lead the deployment of enterprise-grade agentic AI systems within a complex regulated utility environment. This is a hands-on architecture role with broad accountability — you will define agent topology and orchestration patterns, integrate Anthropic Claude and Microsoft Copilot into SAP and enterprise workflows, and own the AWS infrastructure strategy underpinning production AI systems. You will work at the intersection of AI research, enterprise architecture, and operational technology — partnering with SAP functional leads, cloud engineers, compliance teams, and business stakeholders to deliver AI systems that are reliable, explainable, and production-ready at scale. Key ResponsibilitiesAgentic AI Design & Agent Core Architecture Architect multi-agent systems using Agent Core frameworks (LangGraph, AutoGen, or equivalent) — defining agent roles, memory structures, tool registries, and orchestration topologies for enterprise utility use cases. Design and implement ReAct, plan-and-execute, and supervisor-agent patterns for complex, multi-step workflows such as field work order automation, outage root-cause analysis, and regulatory document generation. Define agent evaluation frameworks: success criteria, failure modes, human-in-the-loop checkpoints, and fallback strategies for production agentic pipelines. Establish agent versioning, observability, and audit logging standards to meet regulatory requirements (CPUC, NERC CIP, FERC). Lead proof-of-concept builds for new agent use cases, transitioning validated designs into scalable production architectures. Anthropic Claude Integration Serve as the technical authority for Anthropic Claude deployments across the enterprise — via AWS Bedrock API, direct API, and SDK integrations. Design system prompt architectures, tool-use schemas, and context management strategies for Claude-powered agents operating in regulated data environments. Implement prompt caching, RAG pipelines, and retrieval strategies (OpenSearch, Pinecone) to optimize Claude performance across high-volume utility workflows. Establish guardrails, content filters, and output validation layers that meet enterprise safety and compliance standards. Evaluate and adopt new Claude model releases (Opus, Sonnet, Haiku tiers) and advise on model selection strategy by use case. Microsoft Copilot & M365 Integration Architect enterprise Copilot deployments across M365 — Teams, Outlook, SharePoint, Power Automate — for field operations, customer service, finance, and executive functions. Design and develop Copilot Studio agents and declarative plugins connecting to SAP, GIS, and operational data systems via secure API connectors. Define Copilot governance models: data access scoping, sensitivity label integration, audit logging, and user permission tiers. Lead Copilot extensibility architecture — integrating with Microsoft Graph, Azure AI Search, and enterprise knowledge bases. AWS Cloud Architecture Own the AWS architecture for AI workloads — Bedrock, Lambda, Step Functions, API Gateway, ECS, S3, and VPC security controls. Design multi-region, high-availability agent infrastructure with cost optimization strategies aligned to utility operational budgets. Implement CI/CD pipelines for AI model deployment and agent workflow updates using AWS CodePipeline, CDK, and CloudFormation. Integrate AWS CloudWatch, X-Ray, and OpenTelemetry for end-to-end observability of agent execution traces and LLM call chains. Ensure all cloud architecture meets SOC 2, CPUC data privacy requirements, and NERC CIP cybersecurity standards. SAP Integration & Enterprise Data Architect AI agent integrations with SAP S/4HANA, SAP BTP (Business Technology Platform), and SAP Datasphere — enabling agents to read, write, and act on ERP data in real time. Design secure API and event-driven integration patterns between SAP systems and AI agent pipelines, leveraging SAP Integration Suite and AWS EventBridge. Collaborate with SAP functional leads to identify high-value automation targets: procurement workflows, asset management, customer billing, and workforce dispatch. Implement SAP Joule AI copilot integrations and advise on the SAP AI Core / Generative AI Hub roadmap where applicable. Ensure data lineage, masking, and access control for SAP data flowing into LLM context windows. Architecture Governance & Leadership Define and maintain the enterprise AI architecture reference model — standards, patterns, and guardrails for all AI and agent deployments. Lead architecture review boards for new AI initiatives, ensuring alignment with security, compliance, and scalability requirements. Mentor and guide prompt engineers, ML engineers, and integration developers on agentic design principles. Produce architecture decision records (ADRs), solution blueprints, and executive-level technical narratives. Stay current on the rapidly evolving agent framework landscape (Model Context Protocol, OpenAI Agents SDK, Google ADK) and advise on adoption strategy.  Qualifications Required 8+ years in enterprise architecture, AI/ML engineering, or cloud solutions architecture, with at least 3 years focused on LLM/agent systems in production. Deep hands-on experience with Anthropic Claude — system prompt design, tool use, multi-turn context management, and AWS Bedrock deployment. Proven architecture experience with agentic AI frameworks: LangGraph, AutoGen, CrewAI, LangChain, or equivalent. Strong AWS expertise: Bedrock, Bedrock Agentcore, Lambda, Step Functions, ECS, API Gateway, CloudWatch, IAM, and VPC networking. Experience integrating AI/automation solutions with SAP S/4HANA or SAP BTP in enterprise environments. Microsoft Copilot and Copilot Studio architecture and deployment experience in M365 enterprise tenants. Proficiency in Python and/or TypeScript for agent development, API integration, and infrastructure-as-code. Demonstrated ability to lead technical design across cross-functional teams including security, compliance, and business stakeholders. Preferred Experience in a regulated industry — utilities, energy, financial services, or healthcare. Familiarity with SAP Joule, SAP AI Core, or SAP Generative AI Hub. Knowledge of Model Context Protocol (MCP) and its application in enterprise agent orchestration. AWS Certified Solutions Architect — Professional or AWS Certified Machine Learning Specialty. Microsoft Certified: Azure AI Engineer Associate or Copilot-related certification. Exposure to NERC CIP, CPUC, or FERC regulatory frameworks governing utility data and OT systems. Experience with RAG architectures, vector databases (Pinecone, OpenSearch, pgvector), and knowledge graph integrations. Experience with agentic AI, multimodal systems