Lead Specialist, AI Scientist

Position Overview

The AI Scientist (Expert) leads the most technically complex modelling projects within Pearson's AI Center for Enablement (C4E) and acts as a technical reference point for teams across the business. This is not just a senior individual contributor role: the Expert shapes how AI solutions get designed, influences the decisions of the teams they work with, and improves how the C4E does this work over time.

The role requires someone who combines strong technical depth with the judgement to advise on solution design, the communication skills to be useful to non-specialist stakeholders, and the generosity to make the people around them better. Technical quality matters, but so does what gets left behind: better practices, reusable approaches, and partner teams that are more capable than before the engagement.

Key Responsibilities

• Lead AI modelling projects end-to-end, from problem definition through to a validated solution that a partner team can understand, adopt, and maintain.
• Advise product, data, and engineering teams on AI solution design, including model selection, data requirements, architecture decisions, and the tradeoffs involved in each.
• Serve as a technical reference across the C4E and its partner teams: review approaches, answer hard questions, and provide a grounded second opinion on high-stakes decisions.
• Adapt model designs and methods based on partner feedback and validation results, balancing technical rigour with practical constraints.
• Identify where the team's modelling practices could be stronger and act on it: better evaluation approaches, shared templates, clearer processes.
• Create reusable technical resources such as design patterns, evaluation frameworks, and model cards, and actively facilitate knowledge sharing across disciplines.
• Collaborate with the Responsible AI, Data, Platform, and Security teams, ensuring the right people are involved at the right stage and feeding recurring patterns or gaps back to them.
• Support partner adoption by producing documentation and handover materials that are genuinely usable, and staying involved until teams are confident with what has been built.

Expected Deliverables

• Well-validated models and AI systems, with documentation sufficient for partner teams to understand, adopt, and maintain them.
• Technical design write-ups clear enough for a product engineer or new team member to follow without extensive explanation.
• Concrete improvements to how the C4E works: updated evaluation frameworks, reusable templates, or process changes that raise quality over time.
• Reusable outputs such as patterns, playbooks, and model evaluation templates that reduce duplication across future projects.
• Evidence of genuine partner adoption: teams that can use and build on what has been delivered, not just receive a handover.

Required Qualifications

• Substantial hands-on experience building and shipping ML or deep learning models, including complex projects with real production requirements.
• Strong Python skills and fluency across the ML stack (e.g. PyTorch, Hugging Face), with a solid command of experiment design and rigorous evaluation methodology.
• Demonstrated ability to advise on AI solution design and communicate tradeoffs clearly to both technical and non-technical audiences, including senior stakeholders.
• Track record of adapting technical approaches based on feedback and new evidence.
• Experience working across disciplines (data, product, research, compliance) on AI projects of meaningful scope and complexity.
• Clear, concise technical writing and strong facilitation skills.

Preferred Qualifications / Nice to Have

• Experience with generative AI, including LLM fine-tuning, RAG architectures, prompt engineering, or evaluation of LLM-based systems.
• Familiarity with educational technology, assessment, speech processing, or language learning domains.
• Substantive exposure to responsible AI in practice: working through fairness, bias, or explainability problems on real projects, not just in theory.
• Experience improving how a data science or ML team works, not just individual output.
• Familiarity with MLOps tooling and multi-team AI governance workflows.
• Prior experience in an advisory or enablement role.
  
  #LI-MG1