Agentic AI Engineer

Introduction A career in IBM Consulting is built on long-term client relationships and close collaboration worldwide. You’ll work with leading companies across industries, helping them shape their hybrid cloud and AI journeys. With support from our strategic partners, robust IBM technology, and Red Hat, you’ll have the tools to drive meaningful change and accelerate client impact. At IBM Consulting, curiosity fuels success. You’ll be encouraged to challenge the norm, explore new ideas, and create innovative solutions that deliver real results. Our culture of growth and empathy focuses on your long-term career development while valuing your unique skills and experiences. Your Role And Responsibilities The Agentic AI Engineer (Pro-Code, OpenAI Stack) designs, builds, and ships LLM-based AI agents into production. The role owns multi-agent and single-agent orchestration, exposes data and actions to agents through MCP, and is accountable for the evaluation frameworks and guardrails that keep agent output accurate, grounded, and reliable at scale. Preferred Education Master's Degree Required Technical And Professional Expertise Demonstrable history of deploying and operating LLM agents in production, with examples of scale, reliability, and measurable business impact Multi-agent / single-agent orchestration (planner-executor, tool-routing, handoffs, state management, retries, human-in-the-loop) Model Context Protocol (MCP): building and consuming MCP servers and tools to expose data and actions to agents Output-evaluation frameworks: offline + online evals, golden datasets, LLM-as-judge, regression suites, accuracy/groundedness/hallucination metrics, and guardrails Expert Python; OpenAI API + Codex (or equivalent) / OpenAI Agents SDK; function/tool calling, structured outputs, Responses API, streaming Prompt and context engineering, RAG, embeddings/vector stores API integrations, CI/CD, and observability/tracing Preferred Technical And Professional Experience Latency, cost, and token optimization at production scale Secure secrets handling and production observability tooling Show more Show less