Senior Conversation Designer

Autodesk•AMER - Canada - British Columbia - Offsite/Home, BC

22h•CA$94,000 - CA$137,500

About The Position

Shape the voice of AI at Autodesk. As Senior Conversation Designer on the team behind the Autodesk Assistant, you’ll design the conversations that millions of professionals have with AI while they design and make the world’s buildings, products, and content. It’s a chance to define how an assistant should talk, reason out loud, and recover gracefully when it’s wrong, inside the tools people trust to do their best work. Designers at Autodesk can think across the system, champion design ideas, collaborate with stakeholders, advocate for the customer, and deliver impactful data-driven designs. Our designers use creative and innovative solutions to deliver high-quality experiences that delight our customers. The Autodesk Assistant is how customers will experience AI across our Design and Make platform, embedded inside Autodesk products. In this position you’ll design and refine the conversational and agentic interaction patterns that make the Assistant clear, capable, and trustworthy. You’ll join the platform team responsible for the patterns every Autodesk product team adopts, and because the trust bar for AI here is unusually high, our team stewards the patterns that earn it. At Autodesk we look to hire innovators who reimagine what’s possible, those with diverse backgrounds who elevate our thinking, and leaders who help shape the world and our future. For consideration, candidates must include both a resume + portfolio with password when applying. Light travel, 2-3 times per year, North America.

Requirements

4+ years of experience in conversation design, UX design, or a related field
Solid grounding in human-centered design and conversation design, with a portfolio that shows conversational work you’ve shipped
You write like a designer, treating the words in a conversation as the interface
A point of view on what makes an AI conversation trustworthy, and the craft to back it up
You’ve designed conversational, agentic, or chat-based interaction patterns (for example: cancel/retry, undo, reasoning visibility, citations, or grounding), designing them to be reused across products rather than scripted once
You’ve designed generative AI experiences, including prototyping for edge cases, failure modes, and probabilistic outputs
You’re at home in ambiguity, staying effective when the model is probabilistic, the requirements are evolving, and the right answer is something you have to find.
Hands-on comfort with prompt design and system-prompt architecture as a design material, used to shape how a model responds
A working understanding of how large language models behave, including their capabilities, limits, and failure modes
You’re AI-native in how you work. You use tools like ChatGPT, Claude, Cursor, or Claude Code in your daily practice
Fluency in Figma, plus the comfort to pick up adjacent prototyping tools as the work demands
Excellent communication skills and a habit of close cross-functional collaboration

Nice To Haves

Experience designing for multi-agent workflows or tool-use orchestration
Familiarity with frameworks for evaluating AI experience quality (e.g. Minimum Trustworthiness, Transformative Value, or equivalent quality-gate models)
Experience designing within organizational AI governance principles (Trusted AI, Responsible AI, Ethical AI, or equivalent)
Experience contributing to a design system or shared pattern library on a platform team

Responsibilities

Design and refine the conversational and agentic interaction patterns for the Autodesk Assistant, including reasoning visibility (chain-of-thought), action controls (cancel, retry, accept, undo), and citation and grounding experiences
Design the dialogue flows, conversation architecture, and turn-level behavior that shape how the Assistant responds, clarifies, and follows up
Help shape the Assistant’s voice, tone, and persona, along with the system-prompt behavior that brings it to life consistently across Autodesk products
Design the error, repair, and fallback flows that keep a conversation useful when the model is uncertain or wrong
Partner with product, engineering, AI research, and the wider experience design team to ship conversational capabilities that hold up in real customer workflows
Contribute to the Autodesk Assistant pattern library, adopting shared patterns and helping evolve them rather than building one-off solutions
Define and track the quality signals that tell you a conversation worked (task success, user trust, graceful recovery), and use them to refine the experience