Senior AI Software Engineer
- Toronto, ON, Canada
Full-time
Applications open in a new tab.
Job details
- Job type
- Full-time
- Apply by
- Posting language
- English
- Posted
Original job description
Senior Software AI Engineer - Agentic Platforms
Full-Time, Permanent | Downtown Toronto | Hybrid
Our client, a global enterprise, is building an internal AI platform - a multi-tenant, model-agnostic environment spanning a web app, a desktop agent client, an AI App Store, and a skills/tool ecosystem over the company's knowledge.
They're looking for a Senior AI Engineer to help design and build the agent infrastructure at the core of it: orchestration engines, the desktop agent harness, tool/skill registries, and the policy layers that make it all safe to run at enterprise scale.
What you'll do
- Build and extend multi-agent orchestration (LangGraph, Microsoft Agent Framework, Deep Agents) - stateful graphs, checkpointing, sub-agents, human-in-the-loop, long-running execution.
- Work directly with frontier LLM SDKs at the agent-loop level (Anthropic, OpenAI, Google) - not just chat completions.
- Own the model-agnostic routing layer: provider routing, cost/token accounting, graceful degradation as models churn.
- Build MCP integration, sandboxing/process isolation, and a permission system (ALLOW/ASK/DENY) enforced across runtimes.
- Build retrieval, grounding, and citation pipelines that hold up in production.
- Build eval infrastructure - regression gates, tracing, and diagnosability for agent failures.
What we need
- 7+ years building production software, with recent, substantial work on LLM-based systems used by real users.
- Strong TypeScript (Node.js, async, strict typing) and working proficiency in Python (FastAPI, Pydantic).
- Production depth in LangGraph, Microsoft Agent Framework, or Deep Agents.
- Direct experience with at least one frontier provider SDK at the agent-loop level.
- Hands-on with tool calling, MCP (or comparable), streaming (SSE/WebSocket), structured output.
- Cloud-native delivery: containers, Kubernetes (AKS preferred), CI/CD, observability.
Nice to have
- Google ADK, Semantic Kernel, CrewAI, LlamaIndex, AutoGen, Pydantic AI, Mastra.
- Sandboxing/isolation experience (gVisor, nsjail, Firecracker, Seatbelt, Bubblewrap).
- Electron/desktop app engineering, vector search, regulated-industry experience (insurance, banking, financial services).
More about this job
Responsibilities
Design and build enterprise-scale agent infrastructure, including multi-agent orchestration, desktop agent tooling, registries, model routing, and permission controls. Develop production retrieval and citation pipelines, sandboxed MCP integrations, and evaluation, tracing, and diagnostics systems.
Requirements
Requires 7+ years of production software development experience, including recent substantial work on LLM systems used by real users, strong TypeScript, and working proficiency in Python. Candidates should have production experience with a listed agent framework and direct agent-loop experience with a frontier LLM provider SDK, plus cloud-native delivery skills.
Skills
- TypeScript
- Node.js
- Python
- LangGraph
- Microsoft Agent Framework
- Deep Agents
- LLM Agent Development
- Multi-Agent Orchestration
- Frontier LLM Provider SDKs
- Tool Calling
- Model Routing
- MCP Integration
- Sandboxing and Process Isolation
- Retrieval and Grounding
- Evaluation Infrastructure
- Kubernetes
Visa sponsorship
Not detected in the job text
Categories
- Software
- Technology
- Engineering
- Data & Analytics
Keywords
- Agentic Platforms
- Multi-Tenant AI Platform
- Multi-Agent Orchestration
- LangGraph
- Microsoft Agent Framework
- Deep Agents
- Anthropic
- OpenAI
- TypeScript
- Node.js
- Python
- FastAPI
- Pydantic
- MCP
- Tool Calling
- Streaming
- Server-Sent Events
- WebSocket
- Structured Output
- Kubernetes
- AKS
- CI/CD
- Observability
- Sandboxing
- Vector Search
- Retrieval-Augmented Generation
- Human-in-the-Loop
- Electron
- Regulated Industries