Prabhavathi Pemmasani

Developed and deployed Generative AI applications using Python and Large Language Models (GPT-4, LLaMA, Claude) to automate text generation, summarization, and conversational tasks. Built Retrieval-Augmented Generation (RAG) systems with LangChain, FAISS, and OpenSearch for semantic retrieval. Implemented Guardrails AI, designed FastAPI-based microservices, containerized with Docker, orchestrated with Kubernetes, tracked experiments with MLflow, and evaluated model performance with streaming dashboards. Integrated vector databases for embedding storage, enabled RLHF workflows, and followed Responsible AI practices.

Designed end-to-end ML pipelines on AWS SageMaker using Python, scikit-learn, and XGBoost; developed NLP models with Hugging Face transformers and spaCy for text classification and summarization; built semantic search across SLA, policy, and financial documents using FAISS with early Pinecone integration; implemented MLOps with MLflow, Airflow, and Docker; developed ETL pipelines with PySpark, AWS Glue, and Kafka; deployed models via SageMaker endpoints, Docker, and AWS Lambda; applied SHAP/LIME for explainability and built time-series forecasts with ARIMA/Prophet/Optuna.

Designed scalable data marts; developed complex T-SQL/PL-SQL procedures, SSIS ETL pipelines, and SSRS/Power BI dashboards; optimized database performance and automated data integration with PowerShell; integrated SQL Server and Oracle ecosystems with Azure Dataverse and Dynamics 365; provided architectural guidance on data pipelines and mentored junior analysts.