Yaswanth Thota

Machine Learning Engineer @American Express

US

Engineered and deployed LLM-powered assistant using Hugging Face Transformers, OpenAI API, and LangChain.• Designed retrieval-augmented generation (RAG) pipelines leveraging vector databases (Pinecone, FAISS) and document embeddings to improve accuracy and reduce hallucinations by 35%.• Built scalable MLOps pipelines with MLflow, Docker, and Kubernetes, automating model training, versioning, and deployment across cloud environments (AWS SageMaker, Azure ML).• Developed data preprocessing workflows in Python (Pandas, NumPy, Scikit-learn) for entity recognition, semantic search, and knowledge extraction from structured and unstructured datasets.• Integrated Generative AI APIs with internal systems through RESTful microservices in FastAPI/Flask.• Implemented prompt engineering and fine-tuning of transformer models (GPT, BERT, T5), achieving a 40% improvement in response relevance.• Established model monitoring and governance frameworks using Evidently AI, Prometheus, and Grafana to track drift, latency, and compliance with AI model risk management standards.• Optimized agent productivity workflows by introducing recommendation and summarization features, reducing average call handling time by 25% and improving CSAT scores by 18%.• Collaborated with cross-functional teams under Agile/Scrum methodology, delivering AI solutions aligned with responsible AI, security, and data privacy guidelines.

University of North Texas

Master of Science - MS, Computer Science

Vasireddy Venkatadri International Technological University

Computer Science