Head of Fraud

You'll take the reins on building and owning Carmoola's fraud strategy – from designing and deploying the AI agents that catch fraud across our application flow, to building strong relationships across the UK fraud community so we spot new threats early. You'll have a huge amount of autonomy and responsibility from day one, and the opportunity to define what AI-native fraud prevention at Carmoola really means.

This is a high-impact role for someone who is as comfortable writing and shipping production code as presenting to leadership, thinks commercially about the cost of friction, and knows how to build something from scratch. You won't hand requirements to another team – you'll set the strategy, build the models and agents yourself, deploy them into live decisioning and prove their impact on real losses.

• →Own our fraud strategy - Set the direction across customer, document, dealer and vehicle fraud. Decide what we build versus buy, and balance loss prevention against friction for customers and dealers.
• →Lead delivery across the business - Own the fraud roadmap. Shape requirements with Product and Engineering, work with Credit, Ops and Collections to make controls fit how the business really runs, and keep delivery moving from idea to live impact.
• →Build our fraud identification agents - Design and develop AI agents that weigh evidence across application, document, bank, bureau, dealer and vehicle data, flag suspicious cases and explain why.
• →Develop the models underneath - Build fraud classifiers, anomaly detection, document intelligence and network analysis to uncover linked applications, synthetic identities and dealer collusion.
• →Ship it to production - Own deployment end to end, from tested, version-controlled code and CI/CD to safe releases and live monitoring for drift and performance, with clear governance and human oversight wherever customers are affected.
• →Try to break what we've built - Regularly attack our own journey and controls the way fraudsters would, using the latest tools, from deepfakes and synthetic documents to AI-generated identities and automated application attacks. Find the weak spots before anyone else does, then close them.
• →Turn data into decisions - Deliver fraud analytics and MI that show what we're losing, what we're preventing and why, and report against risk appetite to senior leadership and the board.
• →Build our industry network - Build strong relationships with peer lenders, Cifas, National Hunter, the FLA, UK Finance and law enforcement, and turn shared intelligence on emerging threats into new controls fast.

• Fraud experience in UK lending - Significant experience in fraud prevention or financial crime within UK consumer lending, ideally motor finance, point-of-sale or a fast-growth fintech.
• Hands-on data science - You've personally built fraud or risk models and put them into production, not just specified them. You're expert in Python and SQL, with a strong grounding in machine learning.
• Product and delivery mindset - You think like a product manager as much as a data scientist. You define the problem, prioritise ruthlessly, bring Product, Engineering, Credit, Ops and Collections along with you, and deliver outcomes, not just models.
• Agentic AI know-how - Practical experience building with LLMs and AI agents on real decisions, not just demos, including evaluation, guardrails and governance.
• Production engineering discipline - You're at home with Git, testing, CI/CD, cloud platforms and model monitoring, and you ship code you'd trust with live lending decisions.
• A network you actually use - You're well connected in the UK fraud community. You know who to call when a new threat appears, and people call you.
• Commercial and clear-headed - You know friction has a cost, you design controls proportionate to the risk, and you can explain it equally well to engineers and the board.
• Genuine excitement for Carmoola's mission and the chance to shape the future of how people get on the road.