We are looking for an experienced AI Engineer specializing in Agentic AI, LLM workflows, and Harness Engineering to design, develop, deploy, and scale AI-powered solutions in production environments. The ideal candidate will have hands-on experience delivering production-grade AI applications, particularly involving LLM agents, Retrieval-Augmented Generation (RAG), reusable agent skills, and workflow orchestration. The role requires strong engineering ownership, practical experience with Claude Code/Cowork, and the ability to work directly with enterprise customers and non-technical stakeholders. We prioritize the depth of hands-on AI engineering and successful production delivery over years of experience. Key Responsibilities Design, develop, and deploy production-ready AI-powered applications, agent workflows, and LLM-based solutions. Build and maintain agent harnesses, including control loops, tool integrations, skill surfaces, context and state management, and human-in-the-loop checkpoints. Orchestrate multi-step agent workflows using AWS Step Functions, Amazon Bedrock AgentCore, AWS Lambda, and related AWS services. Develop reusable agent skills, integrate external tools and APIs, and implement MCP-based integrations. Design and optimize RAG pipelines, including document chunking, embeddings, vector search, and retrieval strategies. Implement deterministic workflow controls, policy-based routing, retries, exception handling, timeouts, and loop guards. Establish observability, evaluation, testing, and monitoring practices to ensure the reliability, quality, and scalability of AI workflows. Implement AI safety measures, guardrails, prompt-injection defenses, and human-approval mechanisms. Work directly with enterprise customers and business stakeholders to gather requirements, present solutions, and translate business needs into technical implementations. Own end-to-end solution delivery with minimal supervision, from solution design and development through deployment, monitoring, and continuous improvement. Essential Requirements Proven software engineering experience with hands-on AI engineering solutions successfully deployed to production. Experience delivering at least one production project involving LLM agents, reusable agent skills, or RAG pipelines. Practical experience designing, deploying, and scaling AI-powered production workflows using Claude Code and/or Cowork. Strong understanding of LLM APIs, tool use, function calling, streaming, prompt engineering, and context management. Experience with agent orchestration, workflow design, tool integration, and managing LLM-driven application state. Ability to independently own technical solutions and deliver outcomes with minimal guidance. Experience engaging directly with non-technical stakeholders, gathering requirements, and presenting technical solutions in a clear and structured manner. Strong problem-solving skills and the ability to translate business requirements into reliable, scalable AI solutions. Added Advantages / Good to Have 1. Harness Engineering & AWS Agent Orchestration Excellent understanding of agent harness architecture, including control loops, tools, skills, context management, state handling, and human-in-the-loop checkpoints. Hands-on experience with AWS Step Functions and Amazon States Language (ASL) to orchestrate agent steps, branching, retries, parallel execution, and human-approval gates. Experience implementing task-token-based approval workflows and invoking Amazon Bedrock AgentCore runtimes and AWS Lambda from state machines. Understanding of deterministic versus LLM-driven control, including policy-based routing, exception handling, timeout management, and loop prevention. Practical experience hosting, operating, and scaling agent workloads on Amazon Bedrock AgentCore Runtime and exposing tools through AgentCore Gateway. Experience with agent lifecycle management, runtime invocation patterns, and production scaling. 2. Agent Observability, Evaluation & Guardrails Hands-on experience with AgentCore Observability, Amazon CloudWatch, Datadog, or similar monitoring platforms. Ability to implement tracing, metrics, logging, and per-step audit trails for debugging and monitoring production agent workflows. Experience with AgentCore Evaluations, golden datasets, regression testing, and measuring agent quality before and after releases. Practical knowledge of Amazon Bedrock Guardrails and policy controls for prompt-injection defense, personally identifiable information (PII) protection, and hallucination mitigation. Experience implementing human-in-the-loop review and approval mechanisms for sensitive or high-risk workflows. 3. Agentic AI with Claude Experience designing, deploying, and scaling AI-powered production workflows using Claude Code and/or Cowork. Ability to design and publish reusable skills for consumption by agent workflows. Strong understanding of prompt engineering, system prompts, iterative refinement, and evaluation-driven improvements. Experience building and integrating MCP (Model Context Protocol) servers, including FastMCP, and hosting them on AgentCore or similar platforms. Ability to expose tools and enterprise data through MCP servers and APIs, ensuring agents access data through controlled interfaces rather than directly accessing underlying data sources. Experience with multi-agent coordination, task delegation, and result aggregation using agent-to-agent (A2A) patterns. 4. RAG & LLM Application Engineering Hands-on experience building and optimizing Retrieval-Augmented Generation (RAG) pipelines. Understanding of vector databases, document chunking, embeddings, semantic retrieval, and retrieval optimization. Experience integrating LLMs with enterprise systems, APIs, and external tools. Ability to evaluate and improve retrieval relevance, response quality, reliability, and overall agent performance. Preferred Background Consulting, professional services, or client-facing engineering experience. Experience working directly with enterprise customers on AI solution design, deployment, and scaling. Demonstrated ownership of production AI solutions from initial requirements through deployment and ongoing optimization.