We are looking for an AI engineer for a short term contract (3 months, may be extended). The job will be to build agentic workflows that enhance current business processes.
AI and Agentic Engineering:
- LangGraph: state graphs, checkpointing, interrupt/resume for human-in-the-loop
- A2A protocol: agent-to-agent communication, agent cards, capability discovery
- MCP (Model Context Protocol): agent-to-tool integration
- RAG: retrieval-augmented generation, vector embeddings, knowledge bases
- Open Policy Agent (OPA): Rego-based policy-as-code, deny-by-default gatekeeping
- LiteLLM proxy: model-agnostic routing across OpenAI, Anthropic, Google, open-source
- Prompt versioning and decision audit logs: immutable, append-only, correlation IDs
Full-Stack Development:
- Node.js and Python for agent containers, orchestrators, gatekeepers
- React / Next.js for human-in-the-loop portals and approval dashboards
- Postgres (SQL) for platform schemas, agent schemas, LangGraph state persistence
AI-Accelerated Development:
- Claude Code, OpenCode, Cursor: prototype in hours, not weeks
Behavioral Competencies
- Business fluency, facilitation and discovery craft, trusted partner presence, analytical rigor, change leadership, narrative and knowledge capture.
- Works directly with client stakeholders
The Work
Building a multi-agent AI platform on Azure. The work covers the full stack:
- Shared orchestration foundation: human-in-the-loop portal, monitoring
- Multiple production agents built with LangGraph, each with human approval gates
- Data pipelines connecting agent workflows to external APIs and the client's data warehouse
Works alongside the Lead AI Engineer, a second AI Engineer building agents in parallel, and the client's engineering, data, and business teams. Daily standups with client engineers. Regular facilitation sessions with business stakeholders.
Logistics
- Duration: 3 months starting Oct 19, extendable
- Hours: 40 hours per week
- Location: Canadian resource, Eastern time zone. Regular client-facing meetings required (once per week, on site in the GTA).
