AI Engineer

Southern California

Position summary

We are seeking skilled AI Engineers — at both onshore and offshore levels — to design, build, test, and maintain production AI systems within a complex regulated utility environment. Depending on level, you will work on LLM integrations, agentic pipelines, RAG architectures, AWS infrastructure, Copilot extensions, and SAP data integrations.

This is a hands-on engineering role. You will implement the architectures defined by the AI Architect, build prompt pipelines and agent workflows, develop integrations, and participate in testing, evaluation, and continuous improvement of AI systems across the enterprise.

Key responsibilities

Agentic AI & LLM engineering

• Build and maintain multi-step AI agent workflows using frameworks such as LangGraph, LangChain, AutoGen, or CrewAI for utility-specific use cases including field dispatch automation, outage analysis, and document generation.
• Implement ReAct loops, tool-calling patterns, memory management (short-term and long-term), and structured output validation for production agentic systems.
• Develop and version prompt libraries, system prompts, and few-shot examples aligned to enterprise use cases; conduct regression testing and A/B evaluation of prompt changes.
• Integrate Anthropic Claude (via AWS Bedrock) into agent pipelines — handling context window management, tool-use schemas, streaming responses, and error handling.
• Implement RAG pipelines: document chunking, embedding generation, vector store ingestion (OpenSearch, Pinecone, pgvector), and retrieval-augmented response flows.

AWS infrastructure & MLOps

• Build and maintain AI workload infrastructure on AWS: Bedrock, Lambda, Step Functions, API Gateway, ECS/Fargate, S3, and EventBridge.
• Implement CI/CD pipelines for AI model and agent deployment using AWS CodePipeline, GitHub Actions, and CDK/CloudFormation.
• Set up observability and monitoring for LLM call chains and agent execution traces using CloudWatch, X-Ray, and OpenTelemetry.
• Optimize inference cost and latency through prompt caching, model tier selection (Opus / Sonnet / Haiku), and batching strategies.
• Ensure all AWS deployments meet SOC 2, CPUC, and NERC CIP security and data privacy standards.

Microsoft Copilot & M365 engineering

• Develop Copilot Studio agents, declarative plugins, and Power Automate flows that surface AI capabilities within Teams, Outlook, and SharePoint.
• Build secure connectors linking Copilot to enterprise data sources — SAP, GIS systems, and operational databases — via Microsoft Graph and Azure API Management.
• Implement Copilot governance controls: sensitivity labels, data access scoping, audit logging, and user permission configurations.

SAP integration & enterprise data

• Build integration layers between AI agent pipelines and SAP S/4HANA and SAP BTP systems using REST/OData APIs, SAP Integration Suite, and event-driven patterns.
• Develop data preparation, masking, and transformation pipelines that safely extract SAP operational data for use in LLM context windows.
• Collaborate with SAP functional consultants to understand data models for assets, work orders, billing, and customer records — translating them into AI-consumable schemas.
• Implement SAP Joule integrations and contribute to AI Core / Generative AI Hub configurations where applicable.

Testing, evaluation & documentation

• Build LLM evaluation frameworks: accuracy, hallucination rate, latency, and task completion benchmarks for production agent systems.
• Write unit tests, integration tests, and end-to-end agent workflow tests; maintain test suites in Git with automated execution on PR.
• Document AI systems: architecture decision records, runbooks, API references, and prompt engineering guides for internal knowledge base.
• Participate in architecture reviews, code reviews, and sprint ceremonies as part of an agile delivery team.

Qualifications

Required — all levels

• 3+ years of professional software or AI/ML engineering experience.
• Hands-on experience building with large language models in production — Anthropic Claude preferred.
• Proficiency in Python and/or TypeScript for API development, agent logic, and data pipelines.
• Experience with at least one agentic AI framework: LangChain, LangGraph, AutoGen, CrewAI, or equivalent.
• Familiarity with AWS services: Lambda, S3, API Gateway, and at least one of Bedrock, ECS, or Step Functions.
• Understanding of RAG architecture: chunking, embedding, vector search, and retrieval patterns.
• Strong written communication — able to document systems clearly for both technical and non-technical audiences.

Preferred

• Experience integrating AI systems with SAP S/4HANA or SAP BTP via REST/OData APIs.
• Microsoft Copilot Studio and Power Platform development experience.
• Exposure to regulated industries — utilities, energy, healthcare, or financial services.
• AWS Certified Developer or AWS Certified Machine Learning Specialty certification.
• Familiarity with Model Context Protocol (MCP) for agent tool integration.
• Experience with vector databases: Pinecone, OpenSearch, or pgvector.
• Knowledge of CPUC, NERC CIP, or FERC data governance requirements.

Required Roles

• 3–5 yrs; LLM + AWS hands-on, building toward full-stack AI delivery
• 5–8 yrs; agent frameworks, Bedrock, SAP API integrations, Copilot
• 8–12 yrs; end-to-end agentic systems, regulated-industry production experience
• 12+ yrs; Staff / Lead Engineer technical leadership, cross-team AI system ownership