Job description:

About the Role We are assembling a specialist team focused on AI-driven workflow transformation, and we are looking for a hands-on Agentic AI Engineer to help lead it. This is a software engineering role not a research or model-training position. You will build production agents on top of foundation models, not train models from scratch.

You will design, build, and operate AI agents capable of autonomous decision-making, planning, and tool use to execute complex customer and internal workflows. The work spans agentic AI engineering, full-stack development, and AI-assisted delivery - applied to real product problems in an enterprise environment where reliability, security, and performance are non-negotiable.

This role is for someone who writes the code and leads the work - architecting a solution and shaping technical direction in the morning, and shipping a reliable, observable agent into production in the afternoon.

What You'll Do Design & Build Agentic Systems Design, build, and manage AI agents that plan, reason, make autonomous decisions, and use tools to execute multi-step customer and internal workflows. Implement agent orchestration, multi-step planning, and tool / function calling that connect LLMs to backend services, APIs, and enterprise data systems. Build LLM-powered and agentic applications using structured outputs, RAG (retrieval-augmented generation), memory, and state management. Contribute reusable internal patterns and components - prompts, tools, and agent scaffolding - so agentic features can be built consistently across the product. Reliability, Guardrails & Production Readiness Build guardrails, validation, and human-in-the-loop controls that keep agent behaviour reliable, safe, and auditable. Exercise sound judgment on when an agent should act autonomously versus when deterministic logic, validation, or human approval should drive behaviour. Support production readiness through strong engineering practice: observability, logging, debugging, CI/CD, and operational reliability. Monitoring & performance: instrument agentic systems end-to-end - measuring task success, latency, and cost - and build the dashboards that surface flexibility, reliability, and performance for enterprise stakeholders. Diagnose, optimise, and monitor AI solutions in production, driving issues to closure with a quality-first, secure-by-design mindset. Collaboration & Technical Leadership

Lead delivery across services and product surfaces, integrating agentic capabilities into production systems in a maintainable way. Partner with engineering, product, and design to define agent behaviours, evaluate solutions, and ship user-facing improvements. Improve quality through evaluation, testing, experimentation, and iteration; establish repeatable reliability practices for LLM applications. Document solutions, reference architectures, and runbooks that enable long-term maintainability and clean operational handoff. What You'll Bring

7+ years of professional software engineering experience building and shipping production applications. 3+ years of hands-on experience building LLM-powered or agentic applications (tool / function calling, structured outputs, RAG, or multi-step agent workflows). Proven enterprise delivery: has built and shipped production-grade LLM / agentic implementations in an enterprise environment - end-to-end, from framing through deployment and monitoring. Writes code and leads: a genuine hands-on builder who also takes technical ownership - guiding design decisions, reviewing code, and setting the standard for the work. Technical Skills

Python - strong, production-grade coding ability with solid backend engineering fundamentals. LLM & agentic applications - hands-on experience designing prompts, tool use, structured outputs, and multi-step agent workflows on foundation models (e.g. Anthropic Claude, OpenAI / Azure OpenAI). Microsoft Azure - building and operating AI solutions in the Azure ecosystem (e.g. Azure OpenAI / Azure AI services) as the primary cloud platform. RAG - designing retrieval pipelines: indexing, embeddings, vector stores, and retrieval / contextualisation patterns. Containerisation & orchestration - hands-on with Docker and Kubernetes for packaging and running services reliably. C# or TypeScript / React - competence in at least one, with the ability to work across application layers and modern web surfaces. API-driven systems - building and integrating REST / GraphQL APIs and microservices, working across structured and unstructured data. Engineering Judgment & Ways of Working Able to turn ambiguous product or engineering requirements into structured, deliverable implementation plans. Strong grasp of core agent design patterns: planning, memory and state, tool use, and error handling and recovery. Attention to quality and a builder mindset - comfortable with code reviews, testing, and continuous improvement. Clear written and verbal communication; can explain AI concepts to both engineers and non-technical stakeholders. Tools & Platforms You'll Work With

Our agentic stack is Microsoft-centric. You will design, build, and operate across the full path from user query to governed tool execution:

Azure AI Foundry - our core platform for model orchestration, agent runtime, deployment, and governance (Foundry prompt agents and hosted agents). Microsoft Foundry Agent Service - the orchestration layer that routes requests to approved tools, data agents, and workflows. MCP (Model Context Protocol) servers - designing and exposing reusable custom MCP servers and connectors that securely link agents to approved tools and enterprise data sources. Microsoft 365 & Teams / M365 Copilot - integrating agentic capabilities into the surfaces where users actually work (Teams actions, Outlook, calendar, SharePoint). Custom tool layer - defining and governing the approved tools an agent can call (Microsoft 365, Power BI / Lakehouse data, quality records and sign-offs). Observability stack - OpenTelemetry and Azure Monitor for advanced agent traceability, tool-call monitoring, and reliability; M365 Admin Centre, Power BI Copilot dashboards, and Viva Insights for usage and adoption visibility.

Nice to Have Has used AI coding tools to meaningfully change how they build software - not just experimented with them. Experience shipping AI-powered product features in enterprise or regulated software (e.g. financial services). Familiarity with LLMOps / MLOps / GenAIOps: environment management, evaluation, red-teaming, and incident response for AI deployments. Exposure to MCP (Model Context Protocol) servers, connectors, and reusable integration standards. Interest in cybersecurity, automation, or workflow-heavy systems. How to Apply

Please include the following with your application: A summary of relevant experience building and shipping LLM-powered or agentic applications. Examples of agents, agentic workflows, prompts, or AI features you have personally designed and built - deployed examples strongly preferred over proofs-of-concept. The tools, platforms, and languages you work with regularly (Python, Azure, RAG stack, Docker/Kubernetes, C#/TypeScript/React). Any demos, repositories, or production examples you are able to share.

Job Type: Fixed term contract Work Location: Hybrid remote in Toronto, ON (Toronto District) Pay: CA$85,297.89 - CA$145,000.00 per year