Staff Analyst - Risk and Fraud Systems

What do we do?

Paddle offers digital product companies a completely different approach to their payment infrastructure. Instead of assembling and maintaining a complex stack of payments-related apps and services, we’re a Merchant of Record for our customers. That means we take away 100% of the pain of payment fragmentation. It’s faster, safer, cheaper, and, above all, way better.

We’re backed by investors including KKR, FTV Capital, Kindred, Notion, and 83North and serve over 10,000 software sellers in 245 territories globally.

The Role:

Build the data, decisioning and automated systems that allow Paddle to take more good risk while reducing fraud losses, unnecessary customer friction and manual work.

Paddle’s Merchant of Record model makes risk a critical part of the product and its economics. Seller quality, buyer fraud, payment acceptance, refunds, disputes, reserves and card-scheme exposure cannot be optimised independently. This role will connect those signals, identify where the largest value is being lost, turn risk priorities into working controls and workflows, prove their effect in production, and establish a durable owner once a capability becomes repeatable.

This is not a strategy-only role and it is not ownership of a large operational team. It is a hands-on build role at the intersection of risk, data, product, engineering and operations.

What you'll do:

• Work hands-on with SQL (Snowflake, dbt), Python, statistics and modern AI-assisted tools to explore data, test hypotheses, prototype controls and evaluate results.

• Map the end-to-end seller risk system, its data, decisions, controls, owners and failure points.

• Establish the baseline before building: fraud losses, alert yield, time-to-detect, exposure at detection and analyst effort, with losses reconciled with Finance.

• Own the health of live controls: every rule and check has an owner, an expected volume, a measured yield and alerting when it goes quiet or drifts, and every change is tested and impact-assessed before release.

• Build and maintain a labelled library of confirmed fraud cases and a shared fraud typology: the basis for rules now and for machine-learning models with Data Science later.

• Build a value-ranked portfolio of risk opportunities.

• Translate risk problems into clear product, data and engineering work with measurable acceptance criteria.

• Design seller risk tiers that set payout timing, reserves and limits, giving good sellers more room and tightening controls on riskier ones.

• Backtest rules and risk scores on point-in-time data, and design control experiments, shadow tests and phased roll-outs that measure both loss reduction and legitimate-customer impact.

• Produce decision artefacts that let senior leaders make trade-offs about risk appetite, engineering investment, customer experience and commercial growth.

• Stay with each major capability until it works in production, then establish its long-term owner and move to the next high-value problem.

Expected 6 month outcomes:

• At least one major seller-risk journey has materially improved on agreed loss, approval, speed or manual-effort measures.

• Fraud/Risk Operations spends significantly less time on repetitive screening and more time on investigation, judgement and improvement.

• Stable capabilities have been handed to durable owners, leaving the role free to take on the next high-value problems.

We'd love to hear from you if you:

• Significant experience building or improving merchant- or seller-side fraud detection at a PSP, acquirer, payfac, marketplace or Merchant of Record, and a working knowledge of how bust-out, transaction laundering and self-dealing show up in the data.

• Strong hands-on analytical ability, fluency in SQL (Snowflake, dbt or similar), Python and statistics, plus experience using LLMs on unstructured data such as website content or support tickets, with measured accuracy.

• Experience working with decision flow systems such as Tactile

• A record of turning ambiguous cross-functional problems into systems that were adopted and produced measurable results.

• Experience designing, backtesting and tuning rules and risk scores on point-in-time data, and evaluating them when labels are delayed, partial and adversarial.

• A habit of treating controls as production systems: version control, testing and impact analysis on every change, monitoring that flags when a rule goes quiet or drifts, and an audit trail with human escalation for high-impact actions.

• Able to put risk trade-offs in money (fraud losses, approval rates, customer friction, operating cost and revenue from seller categories declined today) using a method Finance will sign off.

• Good judgement about when a problem needs a rule, a workflow, an experiment, a vendor, a model or an agent—and when it needs none of them.

• Ability to work credibly with investigators, operators, product managers, engineers, data scientists and executives.

• Comfortable challenging inherited assumptions while keeping the people closest to the work involved.

• More interested in solving and shipping than in building an empire.

• Experience in a Merchant of Record, payments, marketplace, subscription or other adversarial transactional environment is useful but not mandatory.

Everyone is welcome at Paddle

At Paddle, we’re committed to removing invisible barriers, both for our customers and within our own teams. We recognise and celebrate that every Paddler is unique and we welcome every individual perspective. As an inclusive employer, we don’t care if, or where, you studied, what you look like or where you’re from. We’re more interested in your craft, curiosity, passion for learning and what you’ll add to our culture. We encourage you to apply even if you don’t match every part of the job ad, especially if you’re part of an underrepresented group.

Please let us know if there’s anything we can do to better supp