Agentic Engineer II

Nexteer Automotive Corporation•Pontiac, MI

7d•Hybrid

About The Position

Nexteer is seeking an Agentic Engineer to join their fast lane agentic project team. This role focuses on building production agent systems that interact with enterprise tools and run in production environments. The selected candidate will contribute to enabling a digital and AI-enabled engineering workflow. This is an 18-month project assignment, followed by a full-time deployment into the engineering organization. The role is an individual contributor position with a hybrid work schedule.

Requirements

Minimum of 1+ years of relevant experience.
Shipped agents to production with real users.
Deep hands-on experience with at least one of Claude Agent SDK / Claude Code, Microsoft Agent Framework, LangGraph, or equivalent, and willingness to work across several.
Experience writing MCP servers.
Fluency in agent vs. harness concepts, having built both agents within existing harnesses and at least one custom harness.
Proficiency in Python or JS/TS.
Experience with LLM APIs.
Experience with REST APIs & integrations.
Experience with Docker / containers and cloud deployment.
Experience with Evals for measuring agent behavior.
Experience with AI-assisted development (Claude Code, Cursor) as a default way of working.

Nice To Haves

RAG / vector search experience.
OpenTelemetry / observability instrumentation for LLM apps.
Identity literacy (agent vs. user identity, OAuth on-behalf-of, scoped tokens).
Workflow automation (n8n / Power Automate).
Experience with local models (e.g. Ollama) and self-hosted infra.

Responsibilities

Design, build, and ship production AI agents across various frameworks (Claude Agent SDK, Microsoft Agent Framework, LangGraph, or custom harnesses).
Build MCP servers to connect agents to enterprise systems (PLM, MES, ALM, data platform).
Design multi-agent workflows, including orchestration, handoffs, and agent-to-agent communication patterns.
Build both the agent (model, instructions, tools, memory) and the harness it runs in (tool dispatch, context management, retries, state).
Run rapid experiments, including hypothesis, prototype, deploy, evaluate, and iterate cycles with clear evaluations.
Embrace a fail-fast approach with short cycles and quick learning.
Evaluate agent frameworks, models, and tools through hands-on use.
Document validated agent, skills, and MCP patterns for a shared use case library to promote reuse.