Artificial Intelligence Engineer

Apply: request@tytantech.com (Subject: Agentic AI Engineer) About the role We’re looking for a hands-on engineer to design, build, and productionize agentic AI solutions across Microsoft’s ecosystem—Copilot Studio agents, Azure AI Foundry Agent Service, Semantic Kernel Agent Framework, and AutoGen—connecting them to enterprise data, workflows, and apps. What you’ll do • Design agentic workflows (reasoning, planning, tool use, multi-agent collaboration) using Copilot Studio’s agent builder and Semantic Kernel agents; publish to M365 channels where relevant. • Build production agents on Azure AI Foundry Agent Service (threads, tool orchestration, safety, identity/observability) and integrate with APIs, databases, and business systems. • Prototype and optimize multi-agent patterns with AutoGen; graduate PoCs to robust services. • Wire agents into Power Automate, Teams/SharePoint, Microsoft 365, and data platforms (Fabric/Azure) for end-to-end workflows. (Copilot Studio can publish agents into Microsoft 365 Copilot experiences.) • Implement guardrails (prompt/data safety, content filters, telemetry) and document patterns, playbooks, and reference implementations. (Azure AI Foundry emphasizes secure, scalable agent runtimes.) Must-have qualifications • 5+ years building AI/automation solutions; 1–2+ years with Microsoft AI stack. • Practical experience with Copilot Studio agents OR Azure AI Foundry Agent Service. • Strong in Python or Node.js, REST/Graph integrations, and LLM tool-calling. • Familiar with Semantic Kernel and/or AutoGen for multi-agent orchestration. • Comfortable shipping secure, monitored services in Azure (Entra, Key Vault, logging). Nice to have • Experience with Fabric/Azure data sources; M365 plugins/connectors; RPA adjacency. • Knowledge of responsible AI practices in Azure AI Foundry. What you’ll showcase • A Copilot Studio agent deployed to M365 (use Teams/SharePoint entry points). • An Agent Service demo (tools, state/thread mgmt, content safety, observability). • A multi-agent AutoGen PoC (planner/solver/tool-user pattern). • An SK Agent sample (skills, memory, plugins).