Jyothi Kolla

Data Scientist @MetLife

Birmingham, AL, US

Developed robust fraud detection models leveraging Logistic Regression and XGBoost on large-scale transactional datasets exceeding 500K daily records, leading to a 15% reduction in false claims and significantly improving financial accuracy and operational efficiency across MetLife’s claims processing unit.• Engineered scalable, real-time ETL pipelines using Apache Spark and Azure Data Factory, optimizing data ingestion, transformation, and load operations, which reduced end-to-end processing time by 40% while maintaining compliance and reliability at production scale.• Integrated cutting-edge NLP capabilities by combining LangChain, Retrieval-Augmented Generation (RAG), and Hugging Face transformers to create intelligent claims document retrieval systems, decreasing average query response time by 25% and enhancing overall claims review workflow.• Fine-tuned state-of-the-art large language models including GPT-2 and BERT within Azure ML to deliver personalized insurance risk scoring and recommendation systems, improving prediction accuracy by 10% and enabling more dynamic customer profiling strategies.• Designed and deployed real-time anomaly detection pipelines using Isolation Forests and Autoencoders within Spark Streaming, enabling early identification of fraudulent patterns and boosting fraud capture rate by 12% under stringent latency SLAs.• Created scalable and secure RESTful microservices using FastAPI and Flask to operationalize ML models in fraud prevention, enabling seamless integration with enterprise APIs and accelerating deployment of real-time scoring systems in live environments.• Automated the machine learning lifecycle from experiment tracking to deployment using MLflow, Terraform, and CI/CD pipelines in Azure DevOps, significantly reducing deployment time and ensuring consistent, version-controlled model releases.

University of Alabama at Birmingham

Master's degree