At Wells Fargo, I architected a LangChain, ChromaDB, and GPT-4 RAG pipeline for semantic search and cited answers across internal documents, reducing review time by 35%. I also fine-tuned BERT and RoBERTa models for domain-specific text classification, improving accuracy by 22% over baseline models.
At HP, I designed CNN-based models to classify blood and urine cell images, improving diagnostic accuracy by 18% over baseline. I also built image-processing and inference pipelines for real-time clinical workflows.
At CVS Health, I developed REST APIs and refactored monolithic modules into microservices, improving API efficiency and reducing peak-load latency. My work spans model training and evaluation, backend services, and AWS deployment.

