Hire Ryan C. - Data Scientist

Senior Data Scientist on State Farm's Claims data science team, leading production applied ML models that predict bodily injury indemnity and reserve outcomes for large-scale claims portfolios. Own end-to-end model development and lifecycle management across problem framing, feature and label validation, CatBoost training/tuning, evaluation, monitoring, retraining, and stakeholder communication for production models served into enterprise claims systems. My work spans first and third-party injury coverages, with models trained on tens of millions of historical claims/payment records and used in recurring production scoring workflows across large claim populations. Recent model development improved test-set MAE significantly compared to prior automated rule-based processes, supporting reserve accuracy, financial planning, and operational decision support. In addition to production modeling, I lead annual retraining and evaluation workflows, build monitoring views for feature drift, train/test/production comparisons, claim-mix/lifecycle progression, and communicate model behavior clearly to partners across data engineering, ML engineering, claims, finance, actuarial, pricing, experimentation, and leadership. I also helped shape an A/B test for a model-driven claim timing insight, including eligibility logic, assignment tracking, SQL-based readouts, and time-to-offer metrics.

Senior Data Scientist and technical lead within OptumRx, building applied ML systems across pharmacy fraud, population health, and claims intelligence. Owned work from ambiguous problem framing through feature strategy, model development, validation, stakeholder alignment, and IT handoff using Python, SQL, PySpark, XGBoost, Node2Vec, MiniSOM, and Tableau across billions of pharmacy and medical claims. Created OptumRx's first ML-driven Fraud, Waste & Abuse pharmacy prioritization model, using XGBoost on labeled historical audit outcomes and behavioral features to score and rank 60K+ pharmacies for review. Engineered large-scale feature pipelines over billions of prescription claims, including clustering and graph-embedding approaches, to help audit teams prioritize higher-signal investigations and reduce low-yield review time. Built a precision-focused population health risk-ranking model using longitudinal medical and pharmacy claims to identify members with elevated likelihood of future claims-documented Opioid Use Dependence risk; partnered with medical directors, pharmacists, legal, behavioral health, and employer stakeholders. Developed Tableau and proof-of-concept dashboards for executive review, translating model outputs into employer-facing insights and decision support. Optimized PySpark pipelines processing billions of records, reducing failures and runtime by 40%. Served as technical lead for a team of 7 data scientists, guiding discovery, feature strategy, model reviews, stakeholder communication, and mentorship while keeping hands-on ownership of core modeling work.

I am currently working as a Data Scientist under the Technology Development Program. In my first rotation, I primarily explored whether or not one of Optum's Data Assets was research ready. This entailed writing proc sql code in SAS and visualizing the data as Pivot Charts. Now I am building predictive models on a Big Data Platform. This is part of our Data Science University program.