Jash Shah

Master's

Data Science

Johnson AND Johnson

Data Scientist

Part-time Contract Fine-tuned Large Language Models (LLMs) using AWS Bedrock to enhance risk prediction accuracy and reducing patient readmission rates by 18%, optimizing clinical decision support systems. • Engineered a semantic search engine over unstructured clinical notes using Retrieval- Augmented Generation (RAG), fine-tuned LLMs and Hugging Face cross-encoders to re- rank results, accelerating critical patient information retrieval for diagnosis. • Leveraged Apache Spark and Databricks to process large volumes of healthcare data from multiple sources, reducing ETL processing time by 40% and enhancing performance for real-time analytics. • Designed and automated ETL pipelines using Apache NiFi and Talend to seamlessly integrate Electronic Health Records (EHR) and Electronic Medical Records (EMR), increasing data freshness and integrity by 25%. • Enforced HIPAA-compliant data governance strategies, including encryption, anonymization and audit trails, to ensure secure handling and privacy of Protected Health Information (PHI) across the data lifecycle. • Consolidated healthcare datasets using AWS S3 for scalable storage and Amazon Redshift for high-performance analytical queries, achieving 99% data availability for downstream reporting and machine learning applications. • Developed predictive dashboards with Power BI and Plotly to provide clinicians with dynamic, risk-stratified insights into patient populations, treatment effectiveness and operational trends. • Managed end-to-end MLOps lifecycle using MLflow, automating model tracking, version control and deployment, which led to a 15% decrease in model deployment cycle time. • Built and deployed RESTful APIs using FastAPI to integrate machine learning models into clinical applications, enabling real-time inference to support physicians with personalized care recommendations. • Ensured end-to-end regulatory compliance with FHIR, HL7 & ICD standards, maintaining interoperability and standardized communication across diverse healthcare data systems. • Applied advanced statistical modeling techniques such as Bayesian inference and multivariate analysis using SAS and SPSS to support clinical research and evaluate patient treatment outcomes.