Gunnapaneni Bhavya Sri

Ai Ml Engineer @AIG

US

Designed and deployed a scalable fraud detection system using XGBoost, Isolation Forest, and TensorFlow, reducing false positives and saving over $1M annually in manual review costs. Automated retraining with SageMaker Pipelines using live data from AWS S3.• Developed an LSTM-based NLP pipeline leveraging SpaCy, Hugging Face Transformers, and OCR (Tesseract) to extract structured data from unstructured medical records, improving claims processing speed by 35%. Deployed as a Dockerized Flask API for real-time inference.• Architected a real-time fraud detection pipeline by integrating Kafka-based live claim data streams with ML models deployed on AWS SageMaker, enabling low-latency fraud scoring and seamless integration with claims processing via RESTful APIs.•Implemented adaptive model retraining pipelines that captured investigator feedback and fraud case outcomes, enhancing model accuracy and responsiveness to evolving fraud patterns.• Built policyholder segmentation models using K-Means and PCA on behavioral data from MySQL, identifying high-risk users and enabling personalized premium recommendations that increased upsell revenue.• Developed Power BI dashboards tracking model performance (precision/recall), business impact, and user adoption, enabling data-driven decisions for C-level executives.• Automated large-scale data processing pipelines with PySpark and AWS Glue to handle 10TB+ of monthly claim data, ensuring GDPR compliance through data masking and AWS KMS encryption.• Enhanced the model lifecycle with monthly retraining pipelines by capturing investigator feedback and flagged cases, automating retraining to improve adaptability and accuracy over time.

Lewis University

Master's degree

Vignan Institute of Technology and Science

Bachelor of Technology - BTech