Project Details

[Return to Previous Page]

Knowledge Retrieval System for Technical Documentation

Company: Morgan Advanced Materials

Major(s):
Primary: CMPSC

Optional: MATSE

Non-Disclosure Agreement: YES

Intellectual Property: YES

Create a prototype intelligent document retrieval system that allows employees to interact with technical knowledge bases through natural language queries while maintaining complete data confidentiality through offline deployment. Background: Our company maintains extensive technical documentation including product specifications, material datasheets, engineering handbooks, training materials, and historical project data. Currently, employees must manually search through multiple sources to find specific information, which is time-consuming and inefficient. An AI-powered system that can intelligently retrieve and synthesize information from these sources would significantly improve productivity while keeping proprietary information secure within our infrastructure. Project Scope: The student team will design and develop a working prototype of an offline AI agent system with the following core capabilities: - Knowledge Base Construction: Develop processes to ingest and convert diverse document formats (PDFs, presentations, videos, emails, text files) into searchable embedding databases stored locally - Intelligent Retrieval System: Implement a RAG-based or similar architecture using open-source LLM models (e.g., via Ollama) that can run on local hardware without internet connectivity - User Interface: Create an intuitive web-based interface (e.g., using Streamlit) allowing users to: i) Select which knowledge bases to query, ii) Submit natural language questions and iii) Receive accurate, cited responses drawn exclusively from provided documentation - Data Analysis Integration: Develop functionality to interpret numerical data queries, translating user requests into appropriate data file operations - System Integration Framework: Document APIs or methods for integrating the system with existing company applications - Documentation: Provide comprehensive usage guides, system architecture documentation, and maintenance procedures Technical Requirements: - Python-based implementation using open-source components - Local deployment capability (no cloud dependencies) - Modular architecture for future expansion - Security considerations for handling confidential information Deliverables: - Functional prototype system with web interface - Source code with clear documentation - User manual and technical documentation - Integration guide for company IT systems - Final presentation demonstrating system capabilities Sponsor Interaction: The team will work closely with Morgan Advanced Materials technical staff through weekly progress meetings, receive sample documentation for testing, and incorporate feedback throughout the development cycle. Company representatives will provide domain expertise and evaluate system performance against real-world use cases. This project offers students hands-on experience with cutting-edge AI technologies, software architecture design, and practical problem-solving for industrial applications while delivering tangible value to the sponsor organization.