MLSAide is a Streamlit-based chatbot designed to assist Microsoft Learn Student Ambassadors (MLSA) with program-related questions. Try out the MLSAide chatbot to get answers to all your program-related questions as a Microsoft Learn Student Ambassador.

This file contains the Streamlit web application for the MLSAide chatbot. It sets up the user interface and manages the chat interaction. Key features include:

• Initializes the chat interface with a title, welcome message, and informative sidebar
• Manages the chat history using Streamlit's session state
• Handles user input through a chat input field
• Displays both user messages and bot responses in a chat-like format
• Uses a cached resource to initialize the answer generation function
• Calls the get_answer_from_kb function to generate responses based on user queries

This file is responsible for setting up and populating the knowledge base. It processes the source material and prepares it for efficient retrieval. Main functions include:

• read_data_from_pdf(): Extracts text content from a specified PDF file
• get_text_chunks(): Splits the extracted text into smaller, manageable chunks
• get_embedding(): Generates embeddings for text chunks using Azure OpenAI's embedding model
• Creates a Qdrant collection with appropriate vector configurations
• insert_data(): Uploads the embedded text chunks to the Qdrant collection
• The main() function orchestrates the entire process of reading, processing, and storing the data

This file contains utility functions for generating responses to user queries. It handles the integration between the user interface, the knowledge base, and the AI model. Key components include:

• Sets up connections to Azure OpenAI and Qdrant using environment variables
• get_answer_from_kb function:
  • Generates an embedding for the user's query
  • Searches the Qdrant database for relevant context based on the query embedding
  • Constructs a detailed prompt combining the query, retrieved context, and system instructions
  • Uses Azure OpenAI to generate a response based on this prompt
  • Applies rules and formatting to ensure appropriate and helpful responses

Together, these files create a system that ingests knowledge from a PDF, stores it in a vector database, and uses it to generate relevant and accurate responses to user queries about the Microsoft Learn Student Ambassadors program.

• Frontend: Streamlit
• Backend: Python
• AI/ML:
  • Azure OpenAI API for embeddings and text generation
  • Qdrant for vector similarity search
• Data Processing:
  • PyPDF2 for PDF text extraction
  • LangChain for text splitting
• Other Libraries:
  • python-dotenv for environment variable management
  • tqdm for progress bars in data processing

MLSAide was created and developed by Mohammed Raza Syed, a Beta Microsoft Learn Student Ambassador who has been an active member of the MLSA INDIA community since 2023. This project was developed to assist fellow students and ambassadors with program-related queries as part of the Microsoft Learn Student Ambassadors program.