Lead Graph Data Scientist - Identity Analytics

USAA•San Antonio, NC

1d•Hybrid

About The Position

USAA is seeking a Lead Graph Data Scientist specializing in Identity Analytics to develop and implement quantitative solutions that enhance the detection and prevention of identity theft, account takeover, and various types of fraud. This role involves creating machine learning models and deploying graph analytics capabilities to protect USAA and its members. The position requires close collaboration with strategy teams, fraud analytics directors, and model users, as well as technology partners to implement a Member Protection graph technology strategy. Key responsibilities include developing and updating identity theft and authentication models, deploying graph databases and techniques to identify criminal networks, generating fraud-dense rings, integrating new data sources, and driving innovation with advanced techniques like graph neural networks. The role also includes mentoring junior staff and fostering a culture of research and development.

Requirements

Bachelor's degree in mathematics, computer science, statistics, economics, finance, actuarial sciences, science and engineering, or other similar quantitative field; OR 4 years of experience in statistics, mathematics, quantitative analytics, or related experience (in addition to the minimum years of experience required) may be substituted in lieu of degree.
8 years of experience in predictive analytics or data analysis.
6 years of experience in training and validating statistical, physical, machine learning, and other advanced analytics models.
4 years of experience in one or more dynamic scripted languages (such as Python, R, etc.) for performing statistical analyses and/or building and scoring AI/ML models.
Expert ability to write code that is easy to follow, well documented, and commented where necessary to explain logic (high code transparency).
Strong experience in querying and preprocessing data from structured and/or unstructured databases using query languages such as SQL, NoSQL, etc.
Strong experience in working with structured, semi-structured, and unstructured data files such as delimited numeric data files, JSON/XML files, and/or text documents, images, etc.
Excellent demonstrated skill in performing ad-hoc analytics using descriptive, diagnostic, and inferential statistics.
Proven ability to assess and articulate regulatory implications and expectations of distinct modeling efforts.
Project management experience that demonstrates the ability to anticipate and appropriately manage project milestones, risks, and impediments.
Demonstrated history of appropriately communicating potential issues that could limit project success or implementation.
Expert level experience with the concepts and technologies associated with classical supervised modeling for prediction such as linear/logistic models, discriminant analysis, support vector machines, decision trees, and ensemble methods such as Random Forests, XGBoost, LightGBM, and CatBoost.
Expert level experience with the concepts and technologies associated with unsupervised modeling such as k-means clustering, hierarchical/agglomerative clustering, nearest-neighbors algorithms, DBSCAN, etc.
Demonstrated experience in guiding and mentoring junior technical staff in business interactions and model building.
Demonstrated ability to communicate ideas with team members and/or business leaders to convey and present very technical information to an audience that may have little or no understanding of technical concepts in data science.
A strong track record of communicating results, insights, and technical solutions to senior executive management (or equivalent).
Extensive technical skills, consulting experience, and business savvy to collaborate with all levels and subject areas within the organization.

Nice To Haves

US military experience through military service or a military spouse/domestic partner
Graduate degree in a quantitative subject area
Over 5 years of experience with model development or other advanced fraud detection algorithms
Over 4 years of experience with graph databases and graph solutions
Experience in fraud/financial crimes model development

Responsibilities

Develop and continuously update internal identity theft and authentication models to mitigate fraud losses and reduce negative member experience from fraud applications, synthetic fraud, and account takeover attempts.
Closely partner with the Strategy team, Director of Fraud Identity Analytics, Director of Fraud Model Management, and model users on model builds and priorities.
Partner with Technology and other key collaborators to deploy a Member Protection graph technology strategy, including vendor selection, business requirements, data needs, and clear use cases spanning financial crimes.
Deploy graph databases and graph techniques to identify criminal networks engaging in fraud, scams, disputes/claims, and AML, improving fraud detection and loss mitigation.
Generate and prioritize fraud-dense rings to mitigate losses and improve Member experience.
Identify and work with technology to integrate new data sources for models and graphs to augment predictive power and improve business performance.
Exports insights to decision systems to enable better fraud targeting and model development efforts.
Drives continuous innovation in modeling efforts including advanced techniques like graph neural networks.
Develops and mentors junior staff, establishing a culture of R&D to augment the day-to-day aspects of the job.
Gathers, interprets, and manipulates sophisticated structured and unstructured data to enable sophisticated analytical solutions for the business.
Leads and conducts sophisticated analytics demonstrating machine learning, simulation, and optimization to deliver business insights and achieve business objectives.
Guides the team selecting the appropriate modeling technique and/or technology with consideration for data limitations, application, and business needs.
Develops and deploys models within the Model Development Control (MDC) and Model Risk Management (MRM) framework.
Composes and peer reviews technical documents for knowledge persistence, risk management, and technical review audiences.
Partners with business leaders from across the organization to proactively identify business needs and propose/recommend analytical and modeling projects to generate business value.
Works with business and analytics leaders to prioritize analytics and highly sophisticated modeling problems/research initiatives.
Leads efforts to build and maintain a robust library of reusable, production-quality algorithms and supporting code to ensure model development and research efforts are transparent and based on highest-quality data.
Assists the team with translating business request(s) into specific analytical questions, implementing analysis and/or modeling, and communicating outcomes to non-technical business colleagues with a focus on business action and recommendations.
Manages project portfolio milestones, risks, and impediments.
Anticipates potential issues that could limit project success or implementation and escalates as needed.
Establishes and maintains standard methodologies for engaging with Data Engineering and IT to deploy production-ready analytical assets consistent with modeling best practices and model risk management standards.
Interacts with internal and external peers and management to maintain expertise and awareness of leading techniques.
Actively seeks opportunities and materials to learn new techniques, technologies, and methodologies.
Serves as a mentor to data scientists in modeling, analytics, computer science, business acumen, and other interpersonal skills.
Participates in enterprise-level efforts to drive the maintenance and transformation of data science technologies and culture.
Ensures risks associated with business activities are effectively identified, measured, monitored, and controlled in accordance with risk and compliance policies and procedures.