What You'll Do

Design, build, and ship agentic systems that ground personalized listening experiences in cultural context and world knowledge, used by hundreds of millions of Spotify users
Develop and maintain pipelines for extracting, structuring, and serving cultural signals at scale, leveraging LLMs and agentic workflows
Partner closely with teams across Personalization to integrate foundational cultural data and tech into new agentic listening experiences
Own components end-to-end — from data pipelines and model training to production serving and monitoring
Design and build evaluation tooling (including LLM-as-judge frameworks and dataset analysis), and run experiments to evaluate the impact of cultural context signals on user experience and engagement
Help define the technical direction of the squad, contributing to architecture decisions

Who You Are

5+ years of experience building and shipping machine learning models end-to-end
Strong foundation in Python (Java and Scala are a plus) and experienced with GCP tools (e.g. Dataflow, BigQuery)
Hands-on experience with LLMs and agent orchestration frameworks (e.g. LangChain, LlamaIndex, Pydantic), building tool-calling agents, RAG, and vector databases
Built and shipped production-scale, data-driven AI/ML systems, ideally in content understanding, knowledge graphs, NLP, MIR, or related domains
Comfortable operating as a 0-to-1 builder — thrive in ambiguous, exploratory spaces and move from idea to experimentation to production with confidence
Care about building inclusive, user-centric products, and think about AI and ML in the context of products and user impact
Worked effectively in collaborative, cross-functional environments
Care deeply about code quality, reliability, and scalability