Conversational Designer

Netflix Netflix · Big Tech · United States · Remote · Product Discovery & Promotion

Netflix is seeking a Conversation Designer to improve the support experience for members and customer service agents by designing conversations for their virtual agent. This role involves creating human-AI multi-turn conversations, crafting prompts, building scalable conversation frameworks, and influencing AI strategy. The ideal candidate has hands-on experience designing for generative or agentic AI, understanding LLM capabilities and limitations, and experience with prompt design and human-in-the-loop AI evaluations.

What you'd actually do

  1. Design human-AI multi-turn conversations across chat and voice, including intent modeling, dialogue flows, interaction patterns, and error handling that provide a delightful customer experience.
  2. Help craft prompts and agentic interaction patterns that shape model reasoning, clarification, recovery, escalation, and handoffs. Define logic, routing rules, and agent boundaries in partnership with an AI vendor.
  3. Build scalable conversation frameworks by designing reusable templates, playbooks, and guidelines for consistent, high-quality AI interactions.
  4. Influence AI strategy by providing actionable recommendations for AI deployment, feature development, and roadmap planning.
  5. Organize and lead research, tests, and establish and manage feedback loops to improve human-AI interactions.

Skills

Required

  • conversation design
  • voice UX
  • designing for generative or agentic AI
  • prompt design
  • human-in-the-loop AI evaluations
  • creation of golden datasets to train conversational models
  • designing conversations for virtual agent
  • intent modeling
  • dialogue flows
  • interaction patterns
  • error handling
  • agentic interaction patterns
  • LLM-driven systems
  • conversation data, logs, and prompt iteration workflows

Nice to have

  • linguistics
  • HCI
  • cognitive science
  • presenting to executive audiences
  • customer support virtual agent

What the JD emphasized

  • hands-on experience designing for generative or agentic AI
  • human-in-the-loop AI evaluations
  • creation of golden datasets to train conversational models
  • Strong understanding of how LLM-driven systems differ from rule-based systems, including capabilities, limitations, ambiguity, and failure modes.

Other signals

  • designing conversations for virtual agent
  • human-AI multi-turn conversations
  • agentic interaction patterns
  • LLM-driven systems