Technical judgement for AI, data, and software projects.
I work with leading AI labs and organisations using AI, helping them evaluate systems, understand technical risks, improve workflows, and make practical decisions.
I am especially interested in problems where technical clarity matters: whether a system is reliable, whether a workflow is worth automating, how to evaluate model behaviour, and how to turn data into usable evidence.
What I offer
I provide research-informed, hands-on technical support for organisations working with AI, data, and software. This may involve reviewing an existing system, designing an evaluation process, writing code, interpreting results, improving a workflow, or helping a team decide what should — and should not — be automated.
My aim is to give clients clear technical judgement: what is working, what is fragile, what is worth building, and what should be approached with caution.
Ways I can help
AI system evaluation
Independent review of AI tools, model outputs, prompts, workflows, rubrics, failure modes, and output quality.
Useful when an organisation needs to know whether a system is reliable enough for real work.
Work with AI labs
Expert judgement, model assessment, data analysis, coding, research assistance, and structured feedback on model behaviour.
Suitable for projects that need careful review rather than surface-level annotation.
Data and research pipelines
Python, dataset cleaning, exploratory analysis, structured reporting, and reproducible workflows.
Good analysis should make the next decision easier, not merely produce more figures.
Software and automation
Coding support, debugging, scripting, technical workflow improvement, and practical automation.
Often the best intervention is a small piece of well-scoped software that removes repeated work.
What clients receive
Written technical review
A clear assessment of a tool, plan, system, dataset, workflow, or proposed AI deployment.
Evaluation design
Rubrics, test cases, review criteria, failure-mode analysis, and practical recommendations.
Working code
Python scripts, analysis notebooks, automation tools, data-processing pipelines, or prototypes.
Who this is for
- AI labs needing careful technical judgement, model evaluation, or research-adjacent support.
- Organisations adopting AI but unsure how to assess quality, safety, or usefulness.
- Teams with data problems that need clear analysis and practical conclusions.
- Charities, churches, research groups, and small organisations needing proportionate technical help.
- Leaders who need an independent technical view before committing money, time, or institutional trust.
How I work
1. Define the problem
We begin with what you are trying to do, what constraints you face, and what a useful outcome would look like.
2. Scope the work
The work is then framed as an advisory call, written review, evaluation design, coding task, analysis project, or ongoing support.
3. Deliver something usable
Outputs are designed to be acted upon: code that runs, reports that can be read, and recommendations that make decisions clearer.
Background
My background includes computer science research, large-scale Internet measurement, routing-state modelling, contactless-payment security, AI evaluation, data analysis, and Python-based software work. I bring both academic depth and practical judgement to technical problems.
Discuss a project
For consultancy enquiries, please send a brief description of the problem, the timescale, and the kind of support you are looking for.