Job Title: Applied AI Engineer
Location: Montreal, QC (Hybrid)
Experience: 3 to 5 years of software engineering, including 1 to 3 years building with generative AI
Role summary:
Build AI agents and LLM applications that run inside the firm's enterprise systems, taking them from first prototype to a service other teams can rely on. You will work alongside senior platform engineers and learn how large regulated systems fit together: how requests are authenticated, how services talk to each other, and how code gets safely to production. Strong engineering fundamentals matter more here than the number of AI frameworks you have used.
What you will do:
- Build agents and LLM workflows using tool calling, RAG and MCP, connected to enterprise systems through their APIs.
- Turn prototypes into services that hold up: clean APIs, tests, logging, error handling and documentation others can run.
- Work within the firm's authentication and entitlement controls when your code calls other systems.
- Package and deploy your work in containers on Kubernetes through the team's CI/CD pipelines.
- Write evaluations for prompts and agents, so changes are measured before they ship.
- Work with business partners to scope use cases and explain trade-offs in plain terms.
- Apply responsible AI, security and data protection controls to everything you build.
Must have:
- 3 to 5 years of software engineering with strong Python.
- Solid fundamentals: data structures, async and concurrency, and networking basics (HTTP, DNS, TLS).
- Built and consumed REST APIs; understand authentication basics such as OAuth 2.0, tokens and API keys, and why services should not share credentials.
- Git, code review, unit and integration testing, and CI/CD pipelines (GitHub Actions, Jenkins or similar).
- Docker, and a working understanding of how applications run on Kubernetes.
- SQL and at least one NoSQL store.
- 1 to 3 years building with LLMs: prompt engineering, RAG, embeddings, vector databases and tool calling.
- Can read and debug code in a larger codebase you did not write, using logs and traces to find problems.
Nice to have:
- Experience at a bank, insurer or other regulated firm, including exposure to change management or audit requirements.
- Hands-on Kubernetes: deploying services, reading pod logs, debugging failed rollouts; Helm.
- Applications in the cloud that talk to on-prem systems, and the networking and identity issues that come with that.
- Enterprise identity: Entra ID (Azure AD), single sign-on, role-based access control.
- Agent frameworks: LangGraph or LangChain; building MCP servers; multi-agent patterns.
- Cloud AI services: Azure AI Foundry, Azure OpenAI or AWS Bedrock, with models such as GPT, Claude or Gemini.
- LLM evaluation and tracing tools: LangSmith, Langfuse, DeepEval or Ragas; OpenTelemetry.
- Messaging and caching: Kafka, Redis.
- Enterprise platforms: ServiceNow, Jira, Confluence, SharePoint, Microsoft Graph, Databricks or Snowflake.