Jainil Kiran Rana

Developed customer retention and lapse-risk prediction models using Python, XGBoost, SQL, and Databricks; engineered behavioral, policy lifecycle, and engagement features from large-scale insurance datasets to improve model performance and business decision-making. Built distributed PySpark feature engineering pipelines for structured and semi-structured policy, claims, and customer interaction data; reduced feature preparation bottlenecks and improved training dataset reliability. Established MLflow-based model lifecycle management and automated CI/CD pipelines for training, validation, deployment, and version control, reducing ML release cycles from 15 days to 4 days and improving deployment consistency. Fine-tuned transformer-based NLP models for policy analysis, correspondence routing, and document classification workflows, achieving accuracy improvements from 84% to 92% across production use cases. Developed retrieval-augmented generation solutions using Azure OpenAI, LangChain, e

Developed machine learning models using Python, Scikit-learn, TensorFlow, and SQL for patient risk stratification, healthcare utilization forecasting, and operational analytics, leveraging feature engineering and ensemble learning techniques across large healthcare datasets. Built Spark-based ingestion and transformation pipelines for claims, patient utilization, and clinical datasets, implementing validation rules, schema enforcement, and feature standardization to support machine learning and analytics workloads. Automated model training, validation, packaging, and deployment workflows using MLflow, Docker, and CI/CD practices, reducing release timelines and improving reproducibility across machine learning environments. Designed NLP solutions for clinical document classification, medical text extraction, and healthcare language processing, reducing manual review effort and improving information accessibility for downstream analytical processes. Developed Kafka and Spark Streaming pi