Agentic AI Engineer

United Contractor Services is building production-grade agentic AI to automate business processes across the organization’s Microsoft and Azure ecosystem. As an on-site Agentic AI Engineer in Kennesaw, GA, you will design, ship, and maintain AI agents that combine tool calling, orchestration, and production governance, with evaluation and observability built in.

You’ll work with leadership and operators to identify high-value automation opportunities, then translate them into delivered agent systems and integrations. The focus is practical engineering: secure identity and authorization, rigorous eval pipelines, and dependable agent operations in real production environments.

Responsibilities

• Design and build production agentic AI systems including tool calling, multi-step reasoning, sub-agent orchestration, memory, retrieval, and human-in-the-loop checkpoints.
• Collaborate with leadership and operators across business units to find automation targets and ship agentic solutions.
• Build and maintain integrations across the existing stack, including Microsoft Graph, Entra, SharePoint, Teams, Procore, Azure Functions, Make.com, and third-party APIs.
• Create evaluation pipelines and observability for production agents, including cost monitoring, quality tracking, failure-mode detection, and regression testing.
• Implement secure agent identity and authorization using managed identities, service principals, OAuth2 flows, least-privilege Graph permissions, and audit logging.
• Establish and uphold AI governance standards across entities, covering data handling, model selection, prompt management, version control, and rollback procedures.
• Document architectures and key decisions so systems remain maintainable beyond individual engagements.
• Stay current with the agentic AI landscape (examples include Claude, OpenAI, Azure AI Foundry, Copilot Studio, LangGraph, MCP) and recommend pragmatic platform approaches.

Requirements

• 3–5 years of professional software engineering experience, including at least 1 year shipping production AI/LLM systems (not prototypes or demos).
• Strong proficiency in Python and/or TypeScript.
• Hands-on experience with at least one major agent framework or SDK, such as Anthropic Claude SDK, OpenAI Agents SDK, LangGraph, Vercel AI SDK, Semantic Kernel, or Azure AI Foundry.
• Real experience building and interpreting eval harnesses to measure whether an agent is working.
• Working knowledge of the Microsoft cloud stack, including Azure Functions, Microsoft Graph API, Entra ID app registrations, OAuth2, and managed identities.
• Backend and integration engineering experience with REST APIs, webhooks, queues, state management, and authentication patterns.
• Strong written communication skills and the ability to explain trade-offs to a technical leader making investment decisions.
• Comfort working onsite in Kennesaw, GA.

Technologies

• Microsoft 365, Entra ID, Intune, Azure Functions, Table Storage
• Entra Domain Services, Procore, Make.com, Microsoft Graph
• Python, TypeScript
• Anthropic Claude SDK, OpenAI Agents SDK, LangGraph, Vercel AI SDK, Semantic Kernel, Azure AI Foundry
• Microsoft Copilot Studio, Power Automate, Model Context Protocol (MCP), SharePoint, Teams
• OAuth2, REST APIs, webhooks, queues, managed identities, service principals
• Bicep, Terraform, GitHub Actions, Azure DevOps pipelines

Benefits

• Competitive base salary commensurate with experience, plus a full benefits package including health, dental, vision, retirement, and paid time off.
• Ongoing learning budgets, conference attendance, and dedicated time for skill development.

What You Bring

• Strongly Preferred: Production experience with Microsoft Copilot Studio, Power Automate, or Azure AI Foundry.
• Strongly Preferred: Experience with Model Context Protocol (MCP) servers or comparable tool-server architectures.
• Strongly Preferred: Familiarity with retrieval-augmented generation (RAG) over enterprise document stores including SharePoint, OneDrive, and Confluence.
• Strongly Preferred: Experience working in a multi-entity or multi-tenant environment with cross-cutting identity and data concerns.
• Strongly Preferred: DevOps/IaC exposure including Bicep, Terraform, GitHub Actions, or Azure DevOps pipelines.
• Strongly Preferred: Familiarity with construction-tech (Procore, Bluebeam) or verticalized SaaS integrations.
• Nice to Have: Cybersecurity or compliance background, particularly around AI governance, data classification, or zero-trust principles.
• Nice to Have: Open-source contributions to agent frameworks or developer tools.
• Nice to Have: Prior experience as the first or second AI hire at an organization.

How You’ll Be Measured

• Time-to-production for new agents and automations.
• Measurable business impact including hours saved, cycle time reduced, error rates lowered, and revenue accelerated.
• System reliability including uptime, eval pass rates, cost efficiency, and incident response.
• Quality of the engineering foundation including maintainability, documentation, security posture, and knowledge transfer.

How to Apply

• Send a resume and a short note describing one agentic AI system you shipped to production.
• Cover letters are not required; the project description is.
• Links to public repos, write-ups, or recorded demos are strongly encouraged.