Director's Note

Stock Sense began as a project I built that aimed to help stock investors stay updated with the most recent financial shifts without wasting hours of their time researching on their own. By using a locally hosted LLM with Ollama and HuggingFaceEmbeddings to vectorize data, Stock Sense uses the practices of Retrieval-Augmented Generation (RAG) to find the most relevant information and summarize it to the user. Additionally, prompt engineering (SMART prompts) was used to help reduce hallucinations that are commonly produced by LLMs and prevent the bot from fabricating incorrect or non-factual information, this ensures that users can rely on Stock Sense to provide accurate and trustworthy information.