Roman S.

Healthcare AI agent for clinical decision support processing EHR data mapped to FHIR standards to identify rare disease cases using LLM-based reasoning. • Designed automated EHR-to-FHIR ETL pipelines ensuring seamless processing and FHIR alignment • Parsed diverse FHIR resources (Patient, Condition, Observation, MedicationRequest, DiagnosticReport) building unified schemas • Implemented clinical data cleaning, deduplication, normalization, and transformation pipelines • Architected Neo4j-based clinical knowledge graph and pgvector semantic search for complex path-based analysis • Resolved data conflict reconciliation across heterogeneous clinical sources • Created visualization layer translating LLM reasoning into actionable dashboards for clinicians • Acted as domain expert bridging clinical guidelines and AI engineering • Ensured data governance and compliance with healthcare standards Tech: Python, Pandas, Azure OpenAI, LangChain, Neo4j, pgvector, FHIR, SNOMED CT, ICD-10, Docker

Pharmaceutical data migration from Quality Management and Clinical Trial Management systems with refactored analytics pipelines. • Collaborated with stakeholders to gather data requirements • Designed ETL pipelines automating data ingestion from Veeva and MSSQL into Snowflake • Developed and optimized DBT data models improving performance and data quality • Built ingestion pipelines using Azure Data Factory for analytical database extraction • Deployed AWS Lambda functions for data ingestion from Veeva • Designed analytical dashboards and operational metrics (CAPA compliance, deviation rates, trial enrollment velocity) • Implemented data quality checks and validation frameworks for clinical and quality systems • Executed unit tests for AWS Lambda functions Tech: Python, SQL, Apache Airflow, Snowflake, Veeva, Power BI, AWS Lambda, Azure Data Factory, DBT, MSSQL, Docker

Data processing pipeline extracting and validating clinical trial data from CTMS with AI-driven quality control. • Designed data extraction pipelines from CTMS automating patient, site, and protocol data retrieval • Built ETL workflows in Python and Apache Airflow handling multi-format data (CSV, Excel, JSON, API) • Designed automated QC rules ensuring completeness, accuracy, and regulatory compliance (GCP, CDISC) • Developed Power BI dashboards for clinical operations translating trial metrics into insights • Implemented clinical data reconciliation logic detecting protocol deviations and adverse event gaps • Integrated validated data into centralized repositories (PostgreSQL, AWS Redshift, S3) • Implemented encryption for PHI data, audit logging, and HIPAA/GDPR compliance • Configured Airflow DAG scheduling and alerting for pipeline monitoring Tech: Python, SQL, AWS, PostgreSQL, Apache Airflow, Power BI, HIPAA