Spotify

Content Designer, Personalization

Shapes the voice and in-product language for Spotify's LLM-powered personalization features, from prompts and system copy to evaluation of generated text.

The role

This individual contributor role sits within Spotify's Personalization Design group, which builds recommendation and discovery experiences such as Discover Weekly and AI DJ. The designer works where language meets generative systems, defining how listeners express intent to the product and how the product talks back. The position spans strategy for agentic features down to hands-on authoring of prompts and interface text.

What you would do

  • Set the strategic direction for how agentic products communicate, including voice, personality, and tonal standards
  • Create shared content frameworks that composition and engineering teams can apply
  • Write, test, and refine prompts and system instructions for generative capabilities
  • Develop reusable prompt scaffolding, trial runs, and safeguards alongside product, engineering, and safety partners
  • Craft accessible, inclusive, and measurable interface copy spanning microcopy, help content, error messaging, onboarding, alerts, and feature naming
  • Translate user and business goals into content needs by joining discovery work, critiques, and localization planning with cross-functional peers
  • Work with engineers to structure human-led assessments of model output and define quality criteria that guide model tuning

What they asked for

  • Experienced content designer or UX writer with a portfolio demonstrating complex problem-solving through language
  • Background in conversation design and character development, plus applying brand voice and tone guidelines
  • Hands-on experience with LLMs, prompt design, and output evaluation
  • Working knowledge of zero-shot versus few-shot prompting and of refining instructions to achieve a target tone while limiting fabricated content
  • Comfortable with ambiguity and able to coordinate across teams and subject areas on flows spanning multiple states, surfaces, or systems

Nice to have

  • Familiarity with LLM-as-a-Judge approaches, scaling evaluations, or light scripting for prompt testing in notebooks
  • Native or near-native fluency in a second language
Selected employer postings. AI-assisted summaries may contain errors. This is not a market-wide sample. Independent of the companies listed.