Architected and deployed an end-to-end chatbot solution, using prompt engineering for open-source LLMs; first by segmenting the types of user questions: if a query relates to a document, it automatically selects the appropriate file, performs OCR, extracts the relevant passages, and formulates the response; if it involves structured data, it generates the appropriate SQL queries on pre-prepared datasets. It can answer all kinds of questions—everything from medication details (like DCI or AMM date), pharmaceutical regulations, to industry news, upcoming events, and even the past annual reports.