Data Scientist - Fraud Decisioning

Liberis are on a mission to unleash the power of small businesses all over the world - delivering the financial products they need to grow through a network of global partners.

At its core, Liberis is a technology-driven company, bridging the gap between finance and small businesses.

Some key info for you:

🌱 We were founded in the UK in 2007

👩🏾 🤝 👨🏼 We are a diverse team made up of over 260 talented people, from more than 27 nationalities!

🌍 We have 6 offices spread across the globe: London, Nottingham, Mumbai, Atlanta, Munich & Stockholm

🎯 We have just been named as one of FinTech’s Finest 50 – by Welcome to the Jungle

💥 We have provided over $3bn of funding to small businesses so far!

The team

We are the Risk Analytics team with a goal to drive intelligent decision-making by applying advanced statistical analytics to a wealth of data. At the heart of the Risk function, our focus is to deliver high-quality fraud management for our customers around the world.

Risk team is a globally team with offices in London, Nottingham and Atlanta US, covers Risk Analytics, Decision Analytics, Fraud Analytics, Underwriting and Collections. We're on a mission to grow Liberis into the world's leading embedded business finance provider, and we're looking for a Fraud Model Developer to help us make that happen!

The role

Are you energised by complex problems, real autonomy, and the chance to innovate? If fraud management - and its constantly changing landscape - excites you, this is the role.

Reporting directly to the Director of Risk Analytics, you’ll use deep data analysis to design, build, and productionise fraud strategies and models across the lifecycle balancing loss reduction with healthy approvals. You’ll work across large, multi-source datasets, run A/B and champion–challenger tests, and turn analytics into clear, deployable decision logic that moves the needle.

What You’ll Be Doing

• Own global fraud decisioning: rules, thresholds, step-up controls optimised for £-EL reduction at stable approval rates.
• Build models end-to-end: problem framing, label/observation window design, sampling, feature engineering, training (logistic/GBM), calibration, back-testing, validation, documentation, and deployment into production decisioning.
• Experiment & ship: A/B and champion–challenger tests;
  cost-based optimisation;
  roll out winners quickly.
• Monitor & govern: Robust dashboards/alerts for model drift, PSI, stability, leakage, review yield, chargeback/refund ratios;
  publish a concise weekly fraud pack.
• Data & vendors: Evaluate new data sources and vendors, integrate where ROI is positive, and track performance over time.
• Cross-functional impact: translate analytics into clear policies/playbooks;
  work with Product/Engineering to land decision logic cleanly and safely.

What We Think You’ll Need

• Experience in an analytical fraud management role with measurable impact (we expect this to be 2-4 years, as a rough guide).
• Up-to-date awareness of emerging fraud trends and the latest controls to manage them with a habit of turning intel into tests, rules, or model features quickly.
• Hands-on modelling experience: feature engineering and building/validating fraud models;
  understanding of ROC/PRcurves, Gini/KS, calibration, stability.
• SQL proficiency for data extraction;
  strong Excel for quick analysis.
• Ability to communicate clearly - turn complex analysis into crisp recommendations.
• Proactive, autonomous working style;
  you know when to dive deep and when to align stakeholders.
• Experience deploying models to production or translating models into rules/strategies in a decision engine.
• Experience with Power BI or Looker for reliable, self-serve dashboards.
• GCP exposure and familiarity with version control (Git) are a plus.
• A solid STEM background helps - but aptitude and impact matter most.

What happens next?

Think this sounds like the right next move for you? Or if you’re not completely confident that you fit our exact criteria, apply anyway and we can arrange a call to see if the role is fit for you. Humility is a wonderful thing, and we are interested in hearing about what you can add to Liberis!

Our hybrid approach

Working together in person helps us move faster, collaborate better, and build a great Liberis culture. Our hybrid working policy requires team members to be in the office at least 3 days a week, but ideally 4 days. At Liberis, we embrace flexibility as a core part of our culture, while also valuing the importance of the time our teams spend together in the office.