Associate Data Scientist (RAG system development and LLM prompt/response evaluation), AtQor
Text
Prompt + Response Writing (SFT)Developed and tuned an end-to-end RAG system for enterprise document querying, including preprocessing and semantic retrieval components. Built LLM pipelines with prompt templates, retrievers, and response generation workflows, then evaluated and adjusted outputs to reduce hallucinations. Implemented performance optimizations such as caching, chunking, and token reduction to improve reliability for downstream AI interactions. • Architected RAG over PDFs, invoices, and compliance reports using Azure Document Intelligence. • Implemented semantic search with Azure AI Search and FAISS embeddings for retrieval. • Performed prompt tuning and response evaluation to improve factuality and accuracy. • Built and deployed FastAPI backend APIs for real-time AI service interaction.