Hima Karan Pusarla

Led migration of on-prem risk models to AWS with Dockerized deployments orchestrated under Kubernetes principles, integrated with MLflow for experiment tracking, and achieved production-grade uptime. Designed and deployed a RAG-powered credit risk intelligence system by integrating financial documents and regulatory filings into vector databases, improving credit risk assessment accuracy. Developed critical credit risk prediction models (XGBoost, Random Forest, Logistic Regression) to enhance high-risk account detection and underwriting policies. Implemented anomaly detection frameworks to proactively identify fraud, reducing false positives. Automated model retraining pipelines with Airflow, ensuring models remain adaptive to evolving credit trends. Applied SHAP for explainability and governance to support regulatory compliance. Created real-time risk dashboards in Tableau for portfolio monitoring.

Architected and automated ETL pipelines using Azure Data Factory, Airflow, and SQL; leveraged Spark/Databricks to accelerate large-scale data processing; designed predictive models (Random Forest, XGBoost, ARIMA, and LSTM) for demand forecasting, churn prediction, and risk evaluation; applied NLP to extract insights from customer feedback; deployed models via Flask and FastAPI REST APIs for real-time forecasting and improved supply chain responsiveness; contributed to business analytics driving gains in sales and market expansion.

Designed and deployed predictive models using Random Forest, XGBoost, Naive Bayes, ARIMA, and LSTM to improve demand forecasting, churn prediction, and risk evaluation, reducing inventory costs and increasing customer retention by 12%. Applied NLP techniques with TensorFlow and PyTorch to analyze customer feedback, extracting insights that improved product strategies and reduced return rates. Built and automated ETL pipelines using Azure Data Factory, Apache Airflow, and SQL, ensuring reliable real-time data ingestion and reducing data latency by 30%. Leveraged Apache Spark and Databricks to process large-scale datasets, achieving a 40% improvement in computation speed for complex analytics workloads. Created Power BI dashboards to visualize demand patterns, customer metrics, and inventory optimization, enabling executives to make faster, data-driven decisions. Integrated ML models into ERP platforms via Flask and FastAPI REST APIs, providing real-time forecasting and enhancing supply