Llm And Rag Developer (Ph.D. Collaboration With Sfsa)
CurrentCollaborated with SFSA to build a Retrieval-Augmented Generation (RAG) pipeline, integrating Large Language Models (LLMs) to enable SFSA members to access and interact with information from their Steel Casting Wiki in natural language. Developed and maintaining the system infrastructure to support scalable and efficient database access, continuously refining response quality based on user feedback and evaluation metrics.Key Contributions:- Replaced traditional keyword search with natural language query capabilities, significantly enhancing user experience in accessing SFSA’s Steel Casting Wiki.- Implementing quality assessment frameworks to improve the accuracy and relevance of RAG responses, ensuring consistent, high-quality information retrieval and generation.Integrated tools like PyTorch, Hugging Face Transformers, LangChain, and Elasticsearch to build a robust, scalable solution for information retrieval.