I'm a Healthcare Data Analyst with a rare combination of clinical depth and data engineering breadth — a licensed Pharm.D. with an M.S. in Health Data Science and 3+ years building production-grade analytics systems across payer, PBM, and hospital settings.

I don't just build reports — I build the infrastructure that makes reporting trustworthy. My work sits at the intersection of SQL engineering, ETL architecture, HEDIS/STAR quality measurement, and ML-powered clinical tools — all grounded in hands-on knowledge of how claims, eligibility, and pharmacy data actually flow in the real world.

🏥 Currently: Sr. Healthcare Data Analyst @ Cardinal Health supporting Medicare and Commercial analytics at scale.

┌─────────────────────────────────────────────────────────────────────┐
│  CLINICAL KNOWLEDGE  +  DATA ENGINEERING  +  ML/ANALYTICS           │
│                                                                       │
│  ✔ Pharm.D.-trained — I understand NDC/GPI drug logic, PDC          │
│    calculation, and medication adherence at a clinical level          │
│                                                                       │
│  ✔ HEDIS/STAR production experience — denominator/numerator/        │
│    exclusion logic, CMS submission-ready, zero audit defects         │
│                                                                       │
│  ✔ SQL engineering at scale — CTE pipelines, window functions,      │
│    query optimization across 3M+ member datasets                     │
│                                                                       │
│  ✔ ML in production — XGBoost risk model actively used by           │
│    Pharm.D. interns for clinical decision support                    │
│                                                                       │
│  ✔ EDI expertise — end-to-end 837/835/834 validation in Facets      │
│    across enrollment, claims, and payment workflows                  │
└─────────────────────────────────────────────────────────────────────┘

Python · XGBoost · SHAP · Streamlit · CDC BRFSS 2019–2021

A multi-disease risk prediction application actively used by Pharm.D. interns in a superspecialty hospital for clinical decision support — not a toy project, a real deployment.

• Predicts patient-level risk for Diabetes, Hypertension, Heart Disease, and Obesity from CDC BRFSS population data
• SHAP explainability surfaces individual risk drivers for clinician interpretability
• Resolved 12 production-level engineering challenges: age bin misalignment, label inconsistencies across survey years, and Streamlit Cloud deployment failures
• Why it matters: Bridges the gap between population health data and point-of-care clinical workflows

Python · XGBoost · SHAP · CMS PDE + Beneficiary Files · 515K+ records (2015–2023)

An end-to-end ML pipeline built on real CMS public data — not synthetic datasets.

• Engineered 14 predictive features including polypharmacy indicators (POLY_5PLUS, POLY_10PLUS), GENERIC_PERCENT, OOP_TOTAL, ESRD_IND
• AUC: 0.974 | Recall: 0.833 for identifying top-10% high-cost Medicare beneficiaries
• SHAP analysis confirmed TOTAL_FILLS as the dominant cost driver (64.4% importance) — aligned with real PBM Medication Therapy Management targeting strategies
• Why it matters: Directly actionable for PBM cost containment and Medicare risk stratification programs

Python · TF-IDF · XGBoost · Clinical NLP · 10,000 SOAP Notes

NLP pipeline classifying unstructured clinical notes into 6 disease categories for automated triage support.

• Built a 150+ term medical abbreviation expansion dictionary to normalize clinical shorthand before vectorization
• TF-IDF (20K features + bigrams) + XGBoost achieved 90% accuracy, macro F1: 0.64 — outperforming Logistic Regression baseline by 27%
• Validated on an external clinical case dataset for out-of-sample generalization
• Why it matters: Demonstrates applied NLP on the messy, abbreviation-heavy text that defines real clinical documentation

SQL Server · PostgreSQL · HEDIS · Claims Analytics · EDI Validation

A production-quality SQL portfolio replicating the logic used at real payers and PBMs.

• Diabetic cohort engine: ICD-10 (E10/E11) + NDC drug pathways with full PDC adherence metrics by drug class
• HEDIS quality measures: Complete 30-day all-cause readmission logic with denominator, numerator, and exclusion specifications
• ETL validation framework: 10 automated data quality rules — financial reconciliation, orphan record detection, eligibility overlap checks, CPT/ICD format validation
• Exception reporting dashboards for data governance and pipeline monitoring
• Why it matters: Shows I can build the foundation that HEDIS and STAR reporting actually depend on

I'm open to Senior Healthcare Data Analyst, Healthcare Analytics Engineer, and Clinical Informatics roles where deep domain knowledge + technical execution both matter.

📧 Saikumarchary1709@gmail.com  |  💼 LinkedIn

"Healthcare data is complex by nature. I build systems that make it simple, reliable, and decision-ready."