AI Scientist

Position Overview

The AI Scientist (Specialist) builds and refines machine learning models within Pearson's AI Center for Enablement (C4E). The C4E is a small team that helps the wider business make good decisions about AI and get things built well. This role sits at the practical end of that work: taking real problems from product, data, and engineering teams and turning them into working prototypes and models.

The Specialist works with a good degree of autonomy on defined problems, collaborates closely with the teams they are building for, and shares what they learn along the way. This is a hands-on role for someone who is comfortable moving between exploration and delivery, and who brings their own technical judgement rather than waiting to be told what to build.

Key Responsibilities

• Implement, fine-tune, and evaluate machine learning models (classic ML and generative AI) based on what each problem actually requires.
• Work closely with product managers, data engineers, and domain experts to understand what teams need, translate that into a clear technical approach, and deliver something concrete.
• Contribute to model design discussions with a clear point of view on tradeoffs: accuracy, latency, cost, and data availability.
• Participate in code reviews as both reviewer and contributor, keeping quality and shared standards consistent across the team.
• Document experiments thoroughly: what was tried, what the results showed, what the limitations were, and what should happen next.
• Share technical knowledge with the team, especially when working with new methods or tools, through write-ups, short sessions, or whatever fits the situation.
• Engage the Responsible AI, Data, and Platform teams early so that solutions meet the right standards before problems accumulate.

Expected Deliverables

• Working prototypes and model implementations that meet what the team actually needed, not just the literal brief.
• Timely delivery with clear communication when something is at risk of slipping.
• Experiment reports covering methodology, results, failure modes, and next steps.
• Documentation clear enough for a teammate to pick up and continue without lengthy hand-holding.
• Contributions to code reviews and team learning that leave shared knowledge better than it was found.

Required Qualifications

• Hands-on experience training, evaluating, and iterating on ML or deep learning models.
• Strong Python skills and familiarity with the standard ML stack (e.g. PyTorch, scikit-learn, Hugging Face).
• Ability to take an ambiguous brief, ask the right clarifying questions, and turn it into a sensible technical plan.
• Able to explain technical decisions clearly to people who are not data scientists, and willing to adapt based on feedback.
• Good instincts for what actually matters in a given problem, with experience using experiment tracking tools (e.g. MLflow, DVC, or equivalent).

Preferred Qualifications / Nice to Have

• Experience with LLMs or generative AI, including fine-tuning, evaluation, or prompt engineering.
• Familiarity with educational technology, assessment, or language learning domains.
• Experience working with unstructured data: text, audio, or multimodal inputs.
• Some grounding in responsible AI principles and how they apply in practice.
• Exposure to cloud ML tooling (AWS, Azure, GCP) or MLOps workflows.
  
  #LI-MG1