Staff Data Scientist, Fraud

Navan•Dallas, TX

62d

About The Position

Navan is expanding its Fraud Risk Management organization to build world-class fraud detection, prevention, and analytics capabilities supporting our rapidly growing travel and expense businesses. We are seeking a highly skilled and visionary Staff / Senior Staff Data Scientist, Fraud, to design and scale advanced data science solutions that protect Navan's customers, platform, and financial ecosystem. This is a strategic and hands-on role, where you'll partner with other fraud strategy members, product, engineering, and data teams to design the ML features, build the rule workflow, build next-generation fraud models, develop actionable insights, and lead the application of AI/ML to mitigate emerging fraud threats both in expense card issuing and travel fraud. The ideal candidate combines deep technical expertise in machine learning and data systems with a strong understanding of payment, identity, and transactional fraud patterns. You'll report to the Head of Fraud Risk Data Science strategy and play a critical role in shaping Navan's end-to-end fraud detection infrastructure and analytical roadmap.

Requirements

10+ years of experience in data science strategy, with a strong focus on fraud detection, risk modeling, or financial crime analytics.
Deep technical expertise in machine learning, predictive modeling, anomaly detection, and network analysis applied to fraud problems.
Proficiency in Python, SQL, and modern ML libraries.
Experience with large-scale data environments such as Snowflake, Databricks, Spark, or equivalent big data platforms.
Strong understanding of payment processing, card networks, identity verification, and behavioral fraud typologies (e.g., synthetic identities, ATO, friendly fraud, first-party fraud, third-party fraud, scams).
Demonstrated success in developing and deploying ML models in production, including monitoring and score drift management.
Proven ability to communicate complex data science findings to technical and non-technical audiences, including executives.
Experience collaborating in cross-functional teams that include Product, Engineering, and Fraud Operations.
Master's or PhD in Computer Science, Statistics, Applied Mathematics, or related quantitative field.

Nice To Haves

Familiarity with fraud detection tools, rule engines, and streaming data systems is a plus.

Responsibilities

Lead the design and deployment of advanced ML features, build rule workflows to detect and prevent fraud across travel and expense.
Lead the design and development of advanced ML and statistical models to detect, predict, and prevent fraudulent behavior across onboarding, payments, and expense workflows.
Drive applied research in anomaly detection, network/graph modeling, real-time clustering, and link analysis to identify emerging fraud patterns and organized fraud rings.
Partner cross-functionally with other Risk Strategy team members, Fraud operations, Engineering, and Product teams to translate insights into scalable prevention rules, thresholds, and model-driven interventions.
Own model lifecycle management from feature engineering and experimentation to monitoring, model retraining, and post-deployment optimization.
Perform root-cause and loss attribution analyses, identifying vulnerabilities and quantifying financial impact to inform control effectiveness and business risk appetite.
Sign up for the stretch fraud loss goals and the ability to drive the roadmap to meet the goal.
Collaborate with Data Engineering and Platform teams to define data schemas, pipelines, and infrastructure that enable real-time fraud monitoring and analytics.
Mentor junior data scientists and fraud analysts, providing technical guidance and driving excellence in experimentation, model governance, and reproducibility.
Contribute to the overall fraud strategy roadmap, helping evolve Navan's machine learning and analytics capabilities to stay ahead of emerging fraud trends.
Partner with external vendors and third-party data sources to enrich detection signals and improve model precision and recall.