General Overview of the Job:
An engineer who designs, builds and operates AI agents on the Microsoft stack and takes them all the way to production. This is a code-first role: agents are written as versioned application code against SDKs and APIs, with the same testing, review and release discipline as any other service. Candidates whose agent experience is limited to low-code or no-code builders are not a match for this profile.
Job Duties and Responsibilities:
- Design and build production AI agents in code on Azure AI Foundry and Azure OpenAI Service, using Python or C# and the relevant SDKs.
- Implement orchestration, tool and function calling, memory and retrieval-augmented generation with Semantic Kernel, the Azure AI Agent SDK or a comparable code framework.
- Integrate agents with Microsoft 365, SharePoint and line-of-business systems through Microsoft Graph and custom APIs, exposed as versioned, testable tools.
- Establish evaluation sets, grounding, guardrails and responsible-AI controls, together with monitoring and cost management in production.
- Work with business teams to turn process pain points into scoped agent use cases, and hand over documented, supportable code.
Required Skills and Proficiency:
- Azure AI Foundry and Azure OpenAI Service
- Agent tooling in code (function calling, Model Context Protocol, custom connectors)
- Agent orchestration frameworks (Semantic Kernel, AutoGen)
- Prompt engineering and retrieval-augmented generation
- Python and C# / .NET
- Azure data services (AI Search, Cosmos DB, Storage)
- Microsoft Graph and line-of-business API integration
- Evaluation, guardrails and responsible AI
- CI/CD and infrastructure as code (GitHub Actions, Bicep or Terraform)
Qualifications and Certifications:
- Bachelor's degree in computer science, engineering or equivalent practical experience.
- Microsoft Certified: Azure AI Engineer Associate; Azure Developer Associate or Azure Solutions Architect Expert is an advantage.
- At least three years of software engineering, including a year or more building large language model or agent solutions that reached production.
