SRI DATTA NADIPOLLA

Lead generative AI and predictive analytics initiatives for enterprise healthcare operations. Built production-ready AI assistants and ML models under model-governance requirements; reduced manual processing time by 40% using predictive models, LLM applications, and RAG pipelines. Improved model performance by 18% via feature engineering, statistical analysis, and hyperparameter tuning for fraud and risk prediction. Accelerated knowledge retrieval using hybrid semantic search with Pinecone/Weaviate/ChromaDB and reduced document-lookup time. Implemented ongoing model monitoring and error analysis to catch performance drift early. Delivered executive visibility through dashboards and technical documentation.

Owned forecasting, personalization, and computer-vision initiatives for automotive sales and quality inspection from data pipeline design through production deployment. Improved demand forecasting accuracy using Python, PySpark, TensorFlow, XGBoost, and LightGBM. Reduced pipeline refresh latency by designing scalable Spark/Databricks pipelines integrating Snowflake, PostgreSQL, MongoDB, and AWS SageMaker. Built retrieval-based recommendation systems with Hugging Face embeddings and vector search. Increased defect-detection precision using computer vision (OpenCV, PyTorch) for damage detection and image classification. Implemented MLOps pipelines (MLflow, Docker, Kubernetes, CI/CD) with automated model monitoring.

Built credit-risk and fraud models supporting regulated lending decisions in collaboration with risk and compliance teams. Improved lending decision accuracy by 15% using credit-risk, loan-default, and fraud-detection models with Python, Spark, and scikit-learn, validated through statistical inference and hypothesis testing. Increased ETL throughput by building distributed pipelines in Spark/SQL and AWS. Strengthened generalization using controlled experiments and cross-validation. Shortened executive reporting cycles via Tableau dashboards for loan performance and delinquency. Supported audit readiness by deploying production-ready models with clear technical documentation.