Come shape the conversational experiences at the heart of Spotify's AI-driven products. You'll design how people talk with Spotify — the flows, the voice, the moment-to-moment interactions that make AI feel natural, useful, and unmistakably Spotify.
This role sits at the intersection of writing and system design. You'll bring sharp editorial instincts and real writing chops, but you'll spend your time designing how conversations work — not producing UI copy. Whether you come from editorial, linguistics, interaction design, or an engineering background with a strong feel for language, you'll have room to experiment, prototype, and define how the next generation of Spotify products talk back.
What you'll do
Design end-to-end conversational experiences — dialogue flows, turn-taking, state and context handling, error and recovery paths, and multi-turn interactions across Spotify's AI products.
Engineer and refine the prompts that drive generative and conversational features, shaping model behavior to produce accurate, reliable, and useful outputs.
Translate product requirements into conversational system design — identifying the capabilities (tools, data, actions, UI affordances) a system needs to fulfil user intent, and defining how the model interprets that intent, retains context, and decides what to do.
Prototype with large language models to probe model behaviors, constraints, and failure modes, and turn what you learn into more capable experiences.
Build evaluation frameworks to measure and improve conversation quality, accuracy, task completion, and user impact.
Partner with product managers, engineers, researchers, designers, and the content design team to bring AI systems into cohesive, human-centered products.
Define guardrails for safe, reliable model behavior across AI-driven interactions.
Help raise the bar for conversation design craft on the team, guiding peers and contributing to shared patterns, evaluation methods, and documentation.
Who you are
Technically capable — comfortable working directly with LLMs and code, writing and running evaluations, using structured formats (JSON, YAML, Python), and diagnosing whether a failure is a prompt problem, a system design problem, or a missing capability.
Practiced at testing and improving conversational systems or prompts through data, feedback, and experimentation.
Energized by ambiguity, creativity, and the challenge of defining new interaction paradigms in a fast-moving AI landscape.
A strong writer and editor with excellent editorial judgment, and a distinct feel for voice and character design — how a system's personality is expressed through prompting and behavior, not just word choice.
A strong cross-functional collaborator who balances technical possibility with creative vision.
Familiar with accessibility, localization, regulatory compliance, and safety in digital experiences.
