Principal AI Scientist (all genders)

Role: Principal AI Scientist Responsibilities: - Design and build conversational AI agents - Develop and integrate RAG pipelines to enrich experiences with fresh, contextual data - Design, tune, and evaluate LLM prompts to improve response quality and user engagement - Integrate LLMs with search, recommendation, and external APIs - Define and implement evaluation frameworks for conversation quality and grounding - Run experiments and optimize assistant performance for business KPIs (conversion, engagement, retention) - Own the architecture, development, and deployment of LLM-powered agent workflows - Collaborate cross-functionally with engineering, product, and data science teams - Mentor colleagues and share best practices in applied AI and agent development Qualifications: - 10+ years of experience in the field of engineering or applied science - Strong Python skills and experience building production-grade applications - Strong analytical and problem-solving skills - Experience optimizing AI systems for business KPIs - Strong plus: Experience building LLM-powered applications or AI agents; prompt engineering, tuning, and evaluation; agent frameworks such as LangChain, LlamaIndex, PydanticAI, DSPy, or similar; integrating LLMs with external APIs, tools, and/or RAG systems; evaluating conversation quality and defining LLM metrics; vector databases, embeddings, or retrieval systems What We Offer: - Employee shares program - 40% off fashion and beauty products sold and shipped by Zalando, 30% off Zalando Lounge, discounts from external partners - 2 paid volunteering days a year - Hybrid working model with 60% (or more) remote per week - Work from abroad for up to 30 working days a year - 27 days of vacation a year to start - Relocation assistance available (subject to prior agreement) - Family services, including counseling and support - Health and wellbeing options (including Gympass) - Mental health support and coaching available - Drive your development through our training platform and biannual peer-to-peer review