AI Engineer at Stripe

Architected and deployed a production-scale LLM-powered document intelligence platform that processes high-volume enterprise documents. Leveraged transformer-based models and Retrieval-Augmented Generation (RAG) to enable contextual semantic search and automated information extraction, reducing manual processing time by 40%. Built end-to-end data preprocessing and feature engineering pipelines using Python, Pandas, NumPy, and SQL, including text preprocessing, tokenization, embeddings, and semantic search. Fine-tuned BERT and integrated GPT-based APIs to enable intelligent summarization and question-answering capabilities. Optimized inference latency through embedding caching and batch processing, reducing response time by 30%. Designed evaluation pipelines using precision, recall, and F1-score to benchmark model performance across document categories. Implemented a RAG pipeline with LangChain and vector databases (FAISS) to enhance contextual accuracy and response relevance. Deployed