Skip to content (Press Enter)

Dallas J Blowers

  • Home
  • AI & Data Projects
    • Human Systems
  • Fabrication Projects
AI & IntelligenceIn Progress

Beaverbot (LLM Multi-Agent System)

Project: BeaverBot – Agentic Maker Assistant & Inventory System

(Capstone Project for Lonely Octopus AI Bootcamp)

The Challenge: Hardware Project Management & Context Switching Developing physical prototypes requires constantly shifting between conceptual design, part sourcing, and inventory management. This context switching breaks flow and leads to poorly documented build logs.

The Solution: A Constrained, Function-Calling AI Agent I engineered BeaverBot, a Streamlit-based AI assistant designed specifically for hardware prototyping environments. Unlike standard LLMs, BeaverBot acts as an integrated system manager. It interacts directly with local databases to read existing project scopes, research necessary electronic components, and autonomously update inventory lists and JSON build logs based on natural language commands.

Key Features & Engineering Focus:

  • Agentic Tool Calling: Designed backend logic allowing the LLM to execute python functions, moving from a passive chatbot to an active database manager (updating inventory tables and logging stubs in real-time).
  • System Guardrails & Safety: Implemented strict domain-prompting constraints to prevent off-topic interactions (e.g., rejecting sports queries), demonstrating an understanding of how to safely deploy LLMs in constrained, task-specific environments.
  • Integrated Workspace UI: Built a multi-page Streamlit interface combining the chat agent, a live Inventory Manager, and a backend system monitor, providing a centralized “mission control” for physical builds.

The Systems Engineering Connection: This project highlights a core competency in Systems Integration. It demonstrates the ability to architect a solution where advanced software (LLMs) interfaces securely with structured data (inventory databases) to support physical product development. The focus on system guardrails and reliable output generation is highly translatable to the software-as-a-medical-device (SaMD) and connected health spaces.

Skills Applied: Python, Streamlit, LLM Integration/Prompt Engineering, API/Database Management, Systems Architecture.

WIP Repo: https://github.com/blowerd/maker-assistant-multiagent

Related Projects

Tug Hill Grant Opportunity Dashboard

March 19, 2026

3D Print Farm Modeling (CTMC)

March 24, 2026

Neural Probe Prototype

February 21, 2026
© Copyright 2026 Dallas J Blowers. Perfect Portfolio | Developed By Rara Theme. Powered by WordPress.