Gen AI Model Implementation for Data Retrieval

This case study presents the implementation of a Gen AI model-based solution by Tiger Analytics for a Fortune 500 biopharmaceutical company aimed at enhancing the data retrieval process for over 3,000 documents. The existing system required extensive manual efforts, prompting the need for a secure solution that could efficiently extract relevant content. The project involved several steps, beginning with scope identification and prioritization, followed by data gathering and exploration. The team utilized Large Language Models (LLMs) and Reinforcement Learning from Human Feedback (RLHF) to create a chatbot interface that provided summarized responses to user queries. Key steps included data pre-processing to resolve incoherent data challenges, LLM selection, and the development of a retriever system to ensure accurate context retrieval. The final model achieved 65% accuracy and reduced query response time by nine seconds, significantly improving user experience by allowing instant access to relevant documents.