Description
Spotify's Subscriptions Mission focuses on converting listeners into lifelong subscribers by delivering seamless, valuable experiences across pricing, packaging, and customer journeys.
The Messaging Platform powers Spotify's communications to over a billion users , from push notifications to emails and in-app messages that connect listeners to the content they love. Within this space, the Paloma squad focuses on message optimization: deciding which message reaches which user, through which channel, and at what moment.
We're evolving how messaging works at Spotify , moving from short-term optimization toward systems that understand long-term user journeys. By combining reinforcement learning approaches with deeper domain signals, we're expanding how machine learning shapes the entire messaging funnel.
Responsibilities
- Design, build, and ship machine learning models that optimize messaging across push, email, and in-app channels
- Plan and run A/B experiments in a multi-objective environment, balancing conversion, engagement, retention, and reachability
- Contribute to reinforcement learning systems that optimize for long-term user outcomes rather than immediate interactions
- Partner with product managers, data scientists, and engineers to define what success looks like and how to measure it
- Own the full ML lifecycle, from data and modeling to deployment, monitoring, and iteration
- Integrate ML models with upstream systems, including domain value signals and opportunity generation frameworks
- Help shape the future of AI-assisted development within the team, exploring how tools can accelerate experimentation and delivery
Requirements
- Strong experience building and deploying machine learning models in production environments at scale
- Comfortable translating business problems into ML solutions and discussing trade-offs with cross-functional partners
- Worked on complex optimization problems such as ranking systems or multi-objective decision-making
- Hands-on experience with PyTorch and distributed systems such as Ray or similar frameworks
- Understand experimentation deeply and can design reliable tests in environments with interacting metrics
- Analyze results using approaches like causal inference or metric decomposition when needed
- Experience with or curiosity about reinforcement learning and long-term optimization systems
- Enjoy working across disciplines and navigating ambiguity while shaping strategy and direction
Where You'll Be
- This role is based in London and Stockholm
- We offer you the flexibility to work where you work best! There will be some in-person meetings, but still allows for flexibility to work from home