Data Science Team Lead, Search & Evaluation

Data Science Team Lead, Search & Evaluation About the team:

Elsevier’s mission is to help researchers, clinicians, and life sciences professionals advance discovery and improve health outcomes through trusted content, data, and analytics. As the landscape of science and healthcare evolves, we are pioneering intelligent discovery experiences — from Scopus AI and Leap Space to Clinical Key AI, Pharma Pendium , and next-generation life sciences platforms. These products leverage retrieval-augmented generation (RAG), semantic search, and generative AI to make knowledge more discoverable, connected, and actionable across disciplines.

We are seeking a Search and Evaluation Data Science Team Lead to join Elsevier’s Platform Data Science organisation — the team driving enterprise-scale AI, retrieval, and evaluation innovation across Elsevier’s global platforms. This role will lead a group of applied scientists advancing lexical, vector, and hybrid retrieval systems; designing robust evaluation frameworks; and shaping the foundation of Elsevier’s next-generation search and AI ecosystem.

This is a unique opportunity to build retrieval and evaluation capabilities that power discovery experiences for millions of users — from researchers accelerating innovation to clinicians making evidence-based decisions.

• Lead and mentor a team of data scientists and applied researchers focused on search, retrieval, and evaluation across Elsevier’s research, life sciences, and health platforms.
• Define and execute the roadmap for enterprise-wide search and retrieval excellence, supporting and developing current and next generation academic and life sciences discovery tools.
• Partner with product, engineering, and data platform leaders to align AI discovery capabilities with researcher, clinician, and pharmaceutical workflows.
• Build a culture of rigorous experimentation, measurable impact, and transparent science, ensuring that all AI-driven retrieval and evaluation work meets Elsevier’s Responsible AI standards.
• Represent Elsevier in cross-functional initiatives shaping the organisation’s retrieval and evaluation strategy at the enterprise level.

• Design and optimise lexical search pipelines for large-scale scholarly, clinical, and biomedical data retrieval.
• Develop and refine vector-based and hybrid architectures using dense embeddings, neural re-ranking, and cross-encoder models to enhance retrieval precision and relevance.
• Advance retrieval-augmented generation (RAG) systems that integrate LLMs with Elsevier’s structured and unstructured data — enabling retrieval-enhanced summarisation, question answering, and content understanding across research and health domains.
• Collaborate on core platform services powering knowledge graphs, semantic enrichment, and generative interfaces that underpin Elsevier’s AI products in science, health, and life sciences.

• Define and own the evaluation framework for retrieval and generative AI systems, combining traditional IR metrics with GenAI-specific measures such as:
• Factual consistency and grounding (alignment of generated responses with retrieved evidence)
• Faithfulness and hallucination rates
• Human-in-the-loop quality ratings
• User engagement and downstream task success
• Build and maintain gold-standard evaluation datasets and annotated corpora across both scientific and biomedical domains.
• Lead offline and online experiments, including A/B testing and reinforcement-driven optimisation for retrieval and generation quality.
• Embed fairness, bias detection, and ethical evaluation into all assessment pipelines, ensuring transparency and trust in Elsevier’s AI systems.

• Collaborate with domain experts, ontology engineers, and biomedical informaticians to integrate scientific taxonomies, citation networks, and clinical ontologies into retrieval systems.
• Incorporate structured data — including datasets, chemical entities, genes, drugs, clinical trials, and patient outcomes — into AI-powered discovery pipelines.
• Advance Elsevier’s knowledge graph and metadata integration…