Note: The job is a remote job and is open to candidates in USA. Infojini Inc is seeking a Principal Agentic AI Engineer to join a centralized Agentic AI team. The role leads production-grade agentic AI development across Microsoft Copilot Studio and Azure AI Foundry while enabling low-code agent development, governance, integrations, testing, observability, and responsible AI practices.
Responsibilities
- Own the pro-code extension layer for Copilot Studio building custom connectors developing Power Platform dataflows and custom actions; and creating reusable component libraries
- Implement advanced Copilot Studio capabilities multi-agent orchestration, Model Context Protocol (MCP) server integration, human-in-the-loop (HITL) approval workflows and Copilot Tuning
- Build and enforce governance standards: lifecycle management, naming conventions, environment management, DLP policies, Entra Agent ID configuration, telemetry and analytics, and cost tracking
- Develop the integration between Copilot Studio and Azure AI Foundry, creating seamless upgrade paths where agents that outgrow Copilot Studio’s capabilities
- Architect and build production-grade custom agentic applications on Azure AI Foundry using Azure OpenAI Service, Azure AI Search, Azure AI Agent Service, Prompt Flow, and Semantic Kernel
- Build and maintain custom RAG pipelines and vector databases and develop agent evaluation and testing infrastructure on Azure: building task-specific benchmarks, LLM-as-judge evaluation pipelines, red-team testing harnesses, and regression test suites
- Implement agent observability, tracing, monitoring, Application Insights, and custom telemetry -providing Agentic AI team with real-time visibility into performance, cost, error rates, and impact
- Contribute to responsible AI practices by implementing guardrails, content safety filters, PII detection and masking, prompt injection defenses, and bias mitigation across all Azure AI Foundry
- Design and deliver training for Copilot Studio, structured in tiers: Foundational (business analysts building first agents), Intermediate (power users adding custom connectors and knowledge sources), and Advanced (technical builders using MCP, HITL, and multi-agent patterns)
- Build and publish “Agent Developer Guide” and an agent template gallery - a curated library of pre-built, governance-approved agent design patterns, connector libraries, governance requirements, testing standards, and deployment checklists
- Track and govern developer adoption metrics: count, usage, business value delivered by decentralized agents, and quality/governance compliance rates—demonstrating the ROI
- Identify when citizen-built agents are hitting the ceiling of low-code capabilities and transition them to pro-code solutions, either by extending with custom connectors or migrating to Azure AI Foundry
- Design the integration layer between agent platforms and enterprise systems of record using custom connectors, APIs, MCP servers, and event-driven architectures
- Architect secure, performant RAG pipelines that ground agents in enterprise knowledge with proper chunking, embedding, indexing, and retrieval strategies
- Establish standards for agent memory, state management, and context handling across both short-lived conversational interactions and long-running autonomous workflows
- Write production-quality code (TypeScript, Python, C#) for custom agents, orchestration logic, tool integrations, evaluation pipelines, and deployment automation on Azure AI Foundry
- Build and maintain CI/CD pipelines for agent development across both Copilot Studio and AI Foundry and develop custom connectors and MCP servers that expose enterprise systems to agents
- Implement comprehensive agent testing: unit tests for tool functions, integration tests for system connectors, evaluation tests using LLM-as-judge and human review, and performance testing
- Define and enforce agent quality standards across both pro-code and citizen-built agents: accuracy, hallucination tolerance, response SLAs, cost, and brand voice
- Implement governance automation for Copilot Studio: DLP policies, Entra Agent ID enforcement, environment controls, agent publishing approval workflows, and automated compliance scanning
- Build agent observability stack: dashboards showing agent health, usage, cost, quality scores, escalation rates, and business impact across all brands and functions
- Own incident response for production agent issues: rapid diagnosis, remediation, root cause analysis, and post-incident improvements
Skills
- 8+ years of software engineering experience (Senior) or 12+ years (Principal), with at least 2 years focused on AI/ML systems, conversational AI, or agentic AI platforms
- Deep hands-on expertise with Copilot Studio, including both low-code authoring (topics, knowledge, actions, agent flows) and pro-code extension (custom connectors, Power Platform components, ALM)
- Strong production experience with Azure AI Foundry (Azure OpenAI Service, Azure AI Search, Prompt Flow, Semantic Kernel, Azure AI Agent Service) building custom agentic applications
- Proficiency in at least two additional agent frameworks beyond the Microsoft ecosystem: LangChain/LangGraph, AutoGen, CrewAI, or comparable orchestration frameworks
- Deep understanding of agentic AI patterns: multi-agent orchestration, tool-use and function calling, RAG architectures, embedding strategies, evaluation methodologies guardrails, and MCP
- Strong software engineering fundamentals: TypeScript, Python, or C#; RESTful and GraphQL API design; event-driven architectures; CI/CD; infrastructure-as-code (Bicep, Terraform); containerization
- Excellent communication skills with the ability to present to both engineering peers and business leaders, translating technical concepts into business-relevant language
- Bachelor's degree in computer science, AI/ML, Engineering, or a related field
- Experience with Sierra AI or comparable AI platforms for consumer-facing customer service agents
- Background in Power Platform administration, governance, and toolkit deployment
- Masters preferred
Company Overview