Senior Data Scientist - Big Data R&D, Identity Graph & KYC

The Big Data R&D team develops cutting-edge big data and graph-based solutions for entity search, entity resolution, and identity matching that power Socure’s KYC and compliance products. As a Senior Data Scientist I, you will lead the design and deployment of advanced ML and graph algorithms on large-scale PII datasets, own end-to-end projects from problem definition through production validation, and serve as a key technical partner to Product, Engineering, and Client-facing teams. You will help define standards for feature engineering, experimentation, and data quality across our identity graph stack, with substantial impact on coverage, accuracy, and fairness.

Responsibilities

Own the design, development, and evaluation of machine learning, statistical, and graph-based algorithms for entity-resolution, identity trust scoring, and anomaly detection on massive datasets; Architect and optimize graph-based identity representations (identity graph structure, linkage rules, clustering) to improve match rates, reduce false positives/negatives, and support downstream fraud and KYC models; Build and maintain scalable data pipelines and feature stores in Spark/PySpark (or Scala), including data normalization, deduplication, and feature computation across large PII datasets in AWS/Databricks environments; Lead A/B tests and offline/online experimentation for new models, features, and data sources; define success metrics, design experiments, and ensure rigorous validation before rollout; Evaluate new internal and external data sources: explore signal quality, design backtests, quantify incremental value, and provide clear recommendations on vendor selection and integration; Partner closely with product managers and engineers to translate ambiguous business and regulatory requirements (e.g., KYC coverage, watchlist matching) into concrete modeling and data roadmaps; Provide deep analytical support to Socure’s compliance and regulatory product suite, including investigative analyses, root-cause analysis for anomalies, and clear narratives for internal and external stakeholders; Contribute to model governance and documentation: clearly explain model logic, data dependencies, limitations, and monitoring plans to internal risk/compliance stakeholders; Mentor junior data scientists and engineers on best practices in data exploration, feature engineering, experimentation, and code quality; Communicate complex technical concepts and trade-offs in a concise, structured way to both technical and non-technical audiences (e.g., product reviews, customer meetings, internal briefings)

Requirements

Master's degree with 3+ years of relevant industry experience, or Ph.D. with 1+ years of experience in applied ML / data science roles; Strong proficiency in Python (preferred) or Scala, including experience with ML libraries such as scikit-learn, XGBoost, TensorFlow or PyTorch; Extensive experience with Spark or PySpark and distributed data systems (e.g., AWS EMR, Databricks) working on very large, messy datasets; Deep understanding of supervised and unsupervised learning, feature engineering, model evaluation, and experiment design (A/B testing, holdout strategies, stratification); Experience developing production-quality data pipelines and automated workflows using Airflow or similar orchestration tools; Solid SQL skills and experience working with large-scale analytical data stores; Demonstrated ability to lead medium-to-large projects end-to-end, make sound trade-off decisions under ambiguity, and influence cross-functional stakeholders with data and clear reasoning

Nice to have

Background in Computer Science, Statistics, Mathematics, or related quantitative fields preferred; Practical familiarity with graph databases and/or graph frameworks (Neo4j, AWS Neptune, GraphFrames, DGL, PyTorch Geometric) and graph algorithms for clustering, link prediction, and community detection is strongly preferred; Experience in at least one of: identity verification, fraud detection, credit risk, or adjacent high-stakes domains is a plus