Senior Research Data Scientist

Joining the AI Strategy, Research & Enablement team, you’ll help shape the future of AI across the organisation. Working at the intersection of cutting-edge research and real-world commercial application, you’ll turn innovative ideas into scalable capabilities that influence products, platforms, and long-term strategy. You’ll collaborate with senior leaders and contribute to decision-making across data science, engineering, and product, helping to prioritise and advance the right AI investments for the future.

We’re looking for a Senior Research Data Scientist to play a key role in developing dunnhumby’s next-generation models across a range of strategic AI initiatives. As part of this newly formed team, you’ll focus on advanced modelling, long-term capability building, and cross-industry collaboration. You’ll design new approaches, build and test prototypes, support engineering teams with productionisation, and contribute to partnerships with academia and external research organisations.

What you’ll be doing:

• Drive the design, experimentation, and evaluation of transformer-based models on retail based challenges.
• Explore innovative architectures (e.g., sequence-to-sequence models, causal transformers, hybrid retrieval + transformer approaches) to deliver step-change improvements in personalisation, recommendations, and forecasting.
• Develop research-grade prototypes and collaborate with engineers to translate them into scalable, production-ready solutions.
• Conduct forward-looking research on generative AI, multimodal learning, and representation learning.
• Identify emerging AI/ML techniques with high potential for retail and consumer analytics, shaping conceptual prototypes and feasibility assessments.
• Drive internal thought leadership by defining standards, sharing knowledge, and advising product, engineering, and data science teams on ML adoption and best practices.
• Work with leading academic groups and private-sector research partners to co-develop new modelling approaches, run joint experiments, or explore cutting-edge AI applications.
• Mentor other data scientists and contribute to growing organisational capability in deep learning and advanced ML.
• Partner closely with internal teams to ensure research pathways are aligned with available tooling and future platform strategy.
• Document best practices and contribute to reusable assets, libraries, and internal frameworks.

What we expect from you:

• Expertise in transformer architectures (e.g., BERT, GPT, T5, Time‑Series Transformers).
• Strong hands-on experience with modern deep learning frameworks (PyTorch preferred).
• Solid grounding in machine learning fundamentals and statistical modelling.
• Familiarity with distributed training and GPU acceleration.
• A track record of delivering impactful models in production or research settings.
• Ability to take ambiguous research ideas and shape them into structured investigations or prototypes.
• Curiosity and a learning mindset — tracking state-of-the-art advancements and evaluating their relevance.
• Strong data manipulation and engineering skills (e.g., PySpark, SQL, cloud-based data tooling).
• Strong communication skills, capable of explaining complex ideas to technical and non-technical audiences.
• Collaborative approach to working with ML engineers, product managers, academic partners, and leadership stakeholders.
• A passion for turning cutting‑edge modelling ideas into practical business impact.
• A mindset that blends scientific curiosity with real-world pragmatism.
• Enthusiasm for shaping the next generation of AI capabilities at dunnhumby, both through hands-on modelling and strategic influence.

Nice to have

• Experience in retail, CPG, recommendations, or customer behaviour modelling.
• Engagement with academic research (publications, workshops, industry collaborations).
• Familiarity with vector databases, embeddings, and retrieval techniques.
• Experience Working on large datasets (efficient loading, batching, and streaming) and related tooling (PySpark preferred).