Senior Software Developer, Mcp/ai

Autodesk Autodesk · Enterprise · Toronto, ON +3

Autodesk is building an agentic AI platform for their Visualization Solutions, specifically expanding their Viewer MCP. This role focuses on owning agentic subsystems end-to-end, including design, implementation, evaluation, and production support. Key responsibilities involve agent orchestration, context retention, RAG, and building evaluation systems for AI behavior, safety, latency, and cost. The role requires strong software development skills with applied ML literacy and experience shipping AI/ML features in production.

What you'd actually do

  1. Design and implement complex agentic subsystems: tool/capability composition, agent Orchestration, context retention across sessions, and retrieval (RAG)
  2. Own subsystems end-to-end — clarify requirements, design, implement, evaluate, deploy, and support in production
  3. Build and own evaluation for AI behavior — quality, safety, latency, and cost — and use it to drive durable improvements
  4. Embed trust into agentic workflows: human-in-the-loop review gates, traceability, reversibility, and confidence signaling
  5. Make sound trade-offs across model choice, prompting/orchestration, and system design, balancing accuracy, latency, cost, and maintainability

Skills

Required

  • TypeScript/JavaScript
  • modern web technologies (e.g. React)
  • delivering production-grade software
  • building AI/ML-powered features (LLMs, agents, RAG)
  • prompt/agent design
  • retrieval
  • evaluating model/agent behavior
  • distributed systems
  • APIs
  • service-oriented architectures
  • automated testing
  • CI/CD
  • Git workflows
  • Agile environments

Nice to have

  • Model Context Protocol (MCP)
  • agent frameworks
  • composable tool/capability architectures
  • vector stores
  • embeddings
  • RAG at production quality
  • Node.js
  • cloud-native architectures (AWS, Docker)
  • observability tooling
  • platform-level or cross-team technical initiatives
  • 2D/3D visualization
  • CAD/BIM
  • graphics technologies (WebGL/WebGPU)

What the JD emphasized

  • agentic AI
  • agentic platform
  • agentic subsystems
  • evaluation for AI behavior
  • trustworthy agentic capabilities
  • production-grade software
  • AI/ML-powered features
  • Applied ML literacy

Other signals

  • agentic AI
  • agentic platform
  • agentic subsystems
  • evaluation for AI behavior
  • trustworthy agentic capabilities