Data Analyst, Sardine

• 3–5 years of experience in data analysis, data science, or a related analytical role — ideally in fraud, risk, fintech, or a data-heavy B2B SaaS environment
• Proficiency in SQL (required) and Python or R for data manipulation, statistical analysis, and visualization
• Solid understanding of evaluation metrics and statistical concepts: precision/recall, AUC/ROC, lift, population distributions, and A/B testing basics
• Experience working with external or third-party datasets — assessing data quality, match rates, and signal value
• Strong written and verbal communication skills; ability to synthesize complex analysis into clear narratives for non-technical stakeholders
• Comfort with ambiguity and the ability to define your own structure in a fast-moving environment

Desirable

• We’re looking for a Data Analyst to join Sardine’s Fraud Intelligence team. This role sits at the intersection of data evaluation, vendor strategy, and fraud detection
• You’ll be the analytical engine behind how we assess, test, and onboard new third-party data signals and vendor partnerships — determining which data assets actually move the needle on fraud outcomes for our clients
• This is a high-ownership, high-visibility role. You’ll work closely with the Head of Fraud, product, data engineering, and client-facing teams to build rigorous testing frameworks and translate raw vendor data into actionable fraud intelligence
• Design and execute structured evaluation frameworks to assess the quality, coverage, and fraud-signal value of incoming data assets from vendor partners
• Build lift analyses, backtests, and champion/challenger comparisons to quantify the incremental value of new data signals against our existing fraud detection stack
• Profile vendor datasets for completeness, freshness, match rates, and population coverage across verticals (crypto, fintech, neobanks, e-commerce, etc.)
• Collaborate with fraud leadership to define evaluation criteria tied to real fraud outcomes — false positive rates, catch rates, precision/recall tradeoffs
• Translate vendor data findings into clear, actionable recommendations: adopt, pilot, deprioritize, or decline
• Partner with data engineering to define ingestion requirements and ensure test environments reflect production-like conditions
• Document evaluation results and maintain an internal knowledge base on vendor data performance over time
• Support ad hoc deep dives into fraud trends, model performance, and client-specific data questions as needed