Help us see potential fairly

The context in three sentences

Merit is a live AI system currently supporting RefuAid’s interest-free lending to refugees rebuilding their lives in the UK. It learns from the way experienced caseworkers recognise potential in people the old scoring systems read as blanks, and every prediction is advisory: a human makes every final decision. Merit AI is a non-profit whose constitution (asset lock, Guardian Share) makes it legally impossible for the organisation to be captured or sold out from under the people who build and back it.

Oversight with real influence

This is an oversight and advisory committee. Merit’s team does the work; the committee brings judgement, challenge and connections. Members will:

• Advise on the model. Offer guidance on Merit’s technical direction: the in-house intelligence layer, the learning loop, new data sources, and the central research problem of learning from human acts of recognition without replacing them.
• Oversee the protections. Hold Merit to its non-negotiables as it grows: no external AI provider, no applicant data leaving Merit’s process, advisory-only predictions, a human in every decision, deterministic compliance grounded in cited regulatory sources.
• Advise on hiring. Give input on the shape of Merit’s technical team and on key hiring decisions, and lend credibility that helps attract exceptional, mission-aligned people.
• Support fundraising. Make introductions where your networks allow, and lend your name and judgement to funding applications and to the planned independent audit and outcome evaluation.

The skills we need across the committee (4 to 6 members)

• AI / machine learning. Production-scale model building; depth in interpretability, human feedback or fairness; appetite for a genuinely novel problem: learning how humans see potential in one another.
• AI ethics. Practical experience turning ethical principles into working governance for deployed AI; you judge whether Merit’s fairness monitoring, human oversight and contestability genuinely protect applicants, not just satisfy a checklist.
• Engineering leadership. You have built and hired small, values-dense technical teams and know how to compete for exceptional talent without equity.
• Finance and consumer credit. FCA and CONC fluency, responsible lending, or senior experience in purpose-driven or community finance.
• Fundraising. You know the funders who back this kind of work, and your name and networks make them lean in.
• Data protection and security. UK GDPR depth, ideally protecting vulnerable populations.
• Lived experience. You have rebuilt a life after displacement or worked alongside those who have; you are the committee’s ground truth for whether Merit’s recognition reads as recognition.

The commitment

• Time: Quarterly meetings (90 minutes, remote-friendly), plus occasional input; roughly one to two days per quarter.
• Term: Initial 12 months, renewable.
• Basis: No equity and no remuneration, in keeping with Merit AI’s asset lock; reasonable out-of-pocket expenses reimbursed.

How to express interest

Send a short note on who you are and which part of Merit’s problem excites you to anna@refuaid.org.