Manhattan, New York, United States
• Led the effort to implement and deploy Artificial Intelligence applications across the firm.
• Deliverable was to provide tools that leverage proprietary internal information, making it accessible, both safely and securely, to internal users.
• Natural language processing (NLP) and large language models were utilized to give applications knowledge of 1,000+ database tables allowing for non-technical users to ask questions of vast amounts of data. This was done by building a multi-agent RAG process that took a question, converted the question to a SQL statement, and then returned the results in the requested format.
• Robust logging, parameter tuning, and a vector database were utilized to ensure that answers were not only complete, but repeatable and properly cited.