Lavanya Sunnapuralla

Led end-to-end development of a customer recommendation module using TensorFlow and PyTorch on AWS SageMaker, integrating collaborative filtering to personalize e-commerce experiences and lift retention by 22%. • Designed and deployed scalable ETL pipelines with Apache Airflow and PySpark on Snowflake, automating data ingestion for ML training sets and cutting processing time by 28%. • Engineered NLP sentiment analysis models via Hugging Face Transformers and spaCy, mining customer feedback to inform UX optimizations that boosted satisfaction scores by 18%. • Built predictive pricing engines with time-series forecasting (Prophet + LSTM in PyTorch), leveraging Databricks for feature engineering and driving 15% profit margin gains across categories. • Orchestrated A/B testing frameworks for ML model variants using MLflow, collaborating with marketing to increase CTR and conversions by 12% on product recommendations. • Implemented MLOps workflows with Kubeflow and Docker for continuous model retraining, enabling real-time dashboards in Power BI that improved operational visibility by 40%. • Developed anomaly detection pipelines with Scikit-learn and Isolation Forest on AWS EC2, flagging fraudulent transactions and reducing manual review errors by 35%. • Mentored 5 engineers on LangChain-based RAG systems for query-driven insights, accelerating analytical workflows and throughput by 20%. • Deployed edge AI models using TensorFlow Lite for mobile personalization, integrating with Kubernetes for low-latency inference and enhancing campaign accuracy by 35%.

Directed full lifecycle of fraud detection module using PyTorch and anomaly detection (Autoencoders), processing transaction data via Azure ML to cut false negatives by 10% and enhance security. • Constructed investment risk forecasting models with XGBoost and LSTM on Databricks, integrating economic indicators to improve portfolio returns by 18% across asset classes. • Pioneered NLP entity extraction pipelines with BERT models and spaCy on Hugging Face, automating regulatory compliance checks and reducing audit exceptions by 14%. • Engineered liquidity forecasting systems using regression ensembles in Scikit-learn and R, deployed via MLflow on Azure to stabilize cash flows with 11% operational gains. • Automated data cleansing and feature engineering pipelines with Alteryx and Trifacta on Snowflake, boosting ML model data reliability by 25% for financial analytics. • Developed interactive portfolio optimization dashboards with Tableau and PyTorch models, enabling real- time reviews that minimized risk exposure and sped up decisions. • Integrated RAG-based NLP pipelines with LangChain and Pinecone vector DB for market trend analysis, optimizing diversification strategies with ML classifiers. • Deployed ROI projection models using PySpark and time-series in Python/R, automating scripts on Airflow to increase review frequency by 30% quarterly. • Led compliance forecasting module end-to-end with Kubeflow CI/CD, combining IBM Watson and Azure ML to streamline KPI tracking and cut reporting time by 27%.