DockBot
My Programing Journey started when I Volunteering in a co-working space called
I used a Python script to process and split PDF documents (theDock Manual), extract their text using Google Document AI, and then chunk and clean the text for embeddings. Each chunk is converted into a vector using Google Cloud's Vertex AI text embedding model. These vectors are stored in a pandas dataFrame, which acts as a vector database. When a user asks a question in Slack, the bot embeds the question, finds the most relevant document chunks using dot product search, and uses Gemini Pro to generate a helpful answer based on the retrieved augmented generation (RAG).
Poll Command
A simple command to create a poll in Slack.
Dockbot creating a poem about the organization
Some fun prompt for Dockbot: create a poem about theDock.