Ingénieur(e) principal(e) en systèmes d’IA | Senior AI Systems Engineer

Matador•Laval, QC

8h•CA$110,000 - CA$130,000•Hybrid

About The Position

This isn’t theoretical work — you’ll design, build, and scale AI-powered products that thousands of businesses rely on every day. You’ll move fast, tackle complex problems, and push the boundaries of what’s possible with multi-agent, multimodal AI systems — and every change you ship will be backed by rigorous evals, not vibes.

Requirements

3+ years of production-grade software engineering experience (backend focus preferred).
Strong Python and TypeScript/JavaScript skills.
Solid experience with MongoDB (or comparable databases), AWS, and modern cloud architecture.
An eval-driven mindset: you measure AI changes against real-world data before shipping them.
Hands-on experience with at least one of: Built LLM agents that coordinate multiple model calls and tools, Built eval pipelines (LLM-as-judge, regression gates) for LLM outputs on real-world datasets, Built agentic retrieval systems (hybrid search, reranking, retrieval that feeds agent context).

Nice To Haves

Experience with the Model Context Protocol (MCP), agent SDKs (Claude Agent SDK, OpenAI Agents SDK, LangGraph), or custom agent harnesses.
Deep knowledge of context engineering, structured outputs, and tool orchestration with LLMs.
Experience with AI guardrails: prompt-injection resistance and PII handling in customer-facing systems.
You’ve shipped a real product and improved it through user feedback.

Responsibilities

Design and build multi-agent LLM systems that coordinate tools, memory, and structured outputs — including agents that retrieve, reason over, and act on data from multiple sources (CRM, inventory, conversations).
Drive eval-driven development: build eval suites (offline benchmarks, LLM-as-judge, regression gates) and the observability and tracing needed to ship prompt and model changes with confidence.
Build and integrate Model Context Protocol (MCP) servers and tools that connect agents to real business systems.
Build realtime voice agents (speech-to-speech models, telephony integration) alongside text-based messaging agents.
Deploy AI workflows into scalable cloud infrastructure (AWS, MongoDB, etc.).
Optimize model routing, cost, and latency across multiple model providers.
Ship and iterate on features end-to-end using Python and TypeScript/JavaScript.