williammayes.com

AI Assurance · LLM Evaluation · Behavioural Science

|

Senior Applied Scientist, Warden AI
Formerly Data Scientist & Senior Researcher, Care Quality Commission, 2022–2026
ONS Data Science Campus Graduate, 2023–2025
University of Surrey, PhD Psychology, 2022

I am a PhD-trained behavioural scientist and data scientist working on independent AI assurance: establishing whether an AI system does what it claims, fairly, and evidencing it to people who need more than a demonstration. I came to it through four years in national healthcare regulation, running bias evaluations of LLM classifications and holding end-to-end ownership of national NHS patient surveys, and before that doctoral neuropsychology research. My path was non-linear by design: clinical psychology settings, experimental neuroscience, and national-scale NHS data work each added a layer that purely technical routes miss. I am drawn to problems where the statistical question and the human question are inseparable. Equally at home designing a psychophysics paradigm, a national patient survey, or an LLM evaluation framework; the common thread is rigorous measurement under real-world constraints.

AI AssuranceBias & Fairness TestingLLM EvaluationBayesian InferenceExperimental DesignPsychometricsCausal InferenceSurvey MethodologyPython · RLLM-as-judge
§1 · Projects
github.com/wpmayes →
§2 · Experience
Full CV →
PeriodRoleContributionMethods

2026–present

Senior Applied Scientist

Warden AI

Independent AI assurance for HR technology: continuous auditing of AI-driven hiring and assessment systems for bias and adverse impact. Work spans counterfactual and subgroup test design, statistical methodology for fairness measurement, and evaluation of LLM-based components, producing findings that have to hold up to enterprise customers, auditors, and regulators.

AI AssuranceBias & Fairness AuditingLLM EvaluationAdverse ImpactPsychometricsStatistical MethodologyPython

2023–2026

Data Scientist / Senior Researcher

Care Quality Commission

Led end-to-end redevelopment of the Children & Young People NHS Patient Survey and held ownership of national NHS patient surveys: sampling strategy, contact design, cognitive testing, question development, QA, and publication. Findings covered by the BMJ, Guardian, Sky News, and BBC. Ran structured bias evaluation of LLM harm classifications across protected characteristics, and led an internal AI governance project cataloguing tool use and drafting risk-based guidelines.

Survey MethodologyBehavioural MeasurementLLM EvalBias TestingXGBoostAzure MLPythonNLPR

2023–2025

Data Science Graduate

ONS Data Science Campus

Competitive cross-government programme: Python, ML, NLP, reproducible analytical pipelines, AWS/Azure, data science ethics. Selected alongside a small cohort from across government departments.

PythonML · NLPRAPAWS · Azure

2022–2023

Researcher

Care Quality Commission

Designed selection weighting methodology for an ethnic minority sample boost; analysis revealed health inequalities masked by broad ethnic category aggregation, shared with NHS England and the cross-government survey research group. Contributed to cognitive testing and question design for the Adult Inpatient Survey. Translated legacy SPSS pipelines into reproducible R and Python workflows; maintained internal team R package.

Survey DesignCognitive TestingSampling & WeightingRRAP

2022

Consultant

Open University

Advised on methodology design for a study examining how movement of people with DCD is perceived by those with ASD; ensured diagnostic criteria alignment with DSM standards; recruited participants through DCD community groups.

Research DesignPsychometrics

2018–2022

Doctoral Researcher

University of Surrey

Four years of original quantitative research: MRS and fMRI neuroimaging, psychophysics, Bayesian computational modelling. Designed experimental paradigms to disentangle bottom-up information processing from top-down executive inhibitory control. First evidence of GABAergic dysregulation in DCD; four first-author peer-reviewed publications.

Bayesian MCMCExperimental DesignCausal InferenceR · MATLABfMRI · MRS
§3 · Press coverage
§4 · Publications
under review

Sensory atypicality in developmental coordination disorder is associated with cortical excitatory-inhibitory imbalance

Mayes, W. P., Gentle, J., Jones, D., Sattar, O., Papadopoulou, A., Ivanova, M., & Violante, I. R.

Journal of Neurodevelopmental Disorders

2024

Audio-visual multisensory integration and haptic perception in adults with developmental coordination disorder

Mayes, W. P., Gentle, J., Ivanova, M., & Violante, I. R.

Human Movement Science · doi:10.1016/j.humov.2024.103180

2023

Exploring Executive Functioning of Adults With Probable Developmental Coordination Disorder Using the Jansari Assessment of Executive Functions

Mayes, W. P., Jansari, A., & Leonard, H. C.

Developmental Neuropsychology · doi:10.1080/87565641.2023.2264424

2022

The Nature and Influence of Sensory Processing Deficits in Developmental Coordination Disorder

Mayes, W. P.

Doctoral thesis, University of Surrey · Thesis ↗

2021

Top-down inhibitory motor control is preserved in adults with developmental coordination disorder

Mayes, W., Gentle, J., Parisi, I., Dixon, L., van Velzen, J., & Violante, I.

Developmental Neuropsychology · doi:10.1080/87565641.2021.1966431