Who we are For decades, insurance software has been built around rigid forms, portals and call centers that create friction for customers and drag for teams. Even as AI has advanced, most insurance work is still manual, slow and spread across disconnected systems.
General Magic builds AI agents that handle quoting, follow-ups, document collection, verification, claims intake and broker communication, connected directly to carrier, MGA and broker systems. We've raised $7.2M USD from Radical Ventures, a16z and several AI luminaries. We reward people who thrive with high autonomy, love shipping quickly and want real ownership. We work onsite in Toronto five days a week.
The role You'll build the AI core of General Magic: the agents and LLM systems behind our product. You own how models are prompted, orchestrated and evaluated, and how they stay quick, cheap and safe in production insurance environments.
In this role, you will Design and build agentic and multi-agent systems with tool use,



state and solid failure handling

Engineer prompts and context deliberately, with guardrails and hallucination controls

Cut cost and latency with concrete levers like caching, context structuring and model selection

Build eval pipelines that catch regressions and tie quality to every release

Work across our backend, APIs and data models to ship features end to end

For this role, you should have A deep, first-principles understanding of how LLMs work, not just API usage

Real agentic systems running in production, not demos

Strong backend fundamentals and the ability to move through unfamiliar codebases quickly

Experience treating evals and safety as core engineering

The drive to ship fast in a high-velocity setting

$120-200k, equity, new MacBook Pro and more.

#J-18808-Ljbffr

📌 AI/ML Engineer (Ontario)
🏢 General Magic
📍 Ontario