We're looking for an Applied AI Engineer to design, ship, and operate the agentic AI systems at the core of Mizo. In this role, you'll take agents from prototype to production — and own how they behave, reason, and recover once real customers depend on them.
Mizo is building AI agents that automate service desk operations for Managed Service Providers. Our agents handle ticket triage, client communication, scheduling, resolutions, and escalation — tasks that traditionally require human intervention. We're growing fast, and on track to become the first fully digital service desk for MSPs. We need someone who can make autonomous agents reliable enough to run unattended.
The ideal candidate lives at the intersection of LLM engineering and production systems. You're as comfortable tuning a prompt and an agent loop as you are reading traces, debugging a failed tool call, and tightening an eval suite. You care about what agents actually do in production, not just what they do in a demo.
Key Responsibilities
- Build production agentic systems: design multi-step agent loops with tool use, planning, memory, and autonomous decision-making that run reliably against real customer workloads.
- Own agent reliability end-to-end: instrument, monitor, and debug agents in production — chase down hallucinations, failed tool calls, runaway loops, and edge cases, then ship fixes.
- Build evals and guardrails: create evaluation suites, regression tests, and safety checks so we can ship agent changes with confidence instead of vibes.
- Optimize cost, latency, and quality: make data-informed tradeoffs across models, prompting strategies, retrieval, and orchestration to keep agents fast, cheap, and accurate at scale.
- Integrate agents with real systems: connect agents to MSP tools, ticketing platforms, and internal APIs so they can take real actions, not just generate text.
- Close the loop with production data: turn real agent runs and user feedback into better prompts, better tools, and a better roadmap.
- Collaborate async-first: communicate clearly in writing, share context proactively, and make decisions with minimal coordination overhead.
Required Qualifications
- Proven experience shipping LLM-powered or agentic features to production — not just prototypes or notebooks.
- Deep, hands-on understanding of agent design: tool use, function calling, planning, memory, retrieval, and multi-step orchestration.
- Strong instincts for evaluation and reliability — you know how to measure whether an agent is actually getting better.
- Proficiency with TypeScript and/or Python, Node.js, and modern web architecture.
- Comfort with ambiguity and rapid iteration. You thrive when the spec is not fully written and the model behavior is non-deterministic.
- Ability to manage your own work independently and ship on time.
- 5+ years of software engineering experience, with meaningful time spent on AI/LLM systems.
Desired Skills and Qualifications
- Experience with agent frameworks, orchestration tools, and LLM observability/tracing stacks.
- Familiarity with building eval pipelines, LLM-as-judge, and prompt/version management at scale.
- Experience with retrieval-augmented generation (RAG), vector stores, and context engineering.
- Familiarity with the MSP, IT services space, or operational workflows in B2B SaaS.
- Experience with Azure cloud services and DevOps workflows.
- Strong async communication skills — you write well and think out loud in docs.
What We Offer
- Competitive salary + ESOP: Real ownership in a company with early traction.
- High autonomy: You decide how to build and operate our agents.
- Real impact: Your agents run in production and handle real customer work every day.
- Montreal office (Plateau): We have an office on the Plateau. You're welcome to come in whenever you want.
- Growth trajectory: Join at the ground floor of a fast-growing AI company.
- Cutting-edge AI: Work with agentic AI, LLMs, and autonomous workflows daily.
Why Mizo?
MSPs are drowning in repetitive service desk work. We're building AI agents that handle it autonomously — and the market is responding. With customers on 4 continents and growing fast, we're proving that agentic AI can transform how IT services are delivered.
This is a rare chance to take agentic AI all the way to production at scale, and own the systems that make it trustworthy. You'll work directly with the founding team, see your agents handle real tickets within days, and have a direct line from production data to the next thing you build.