top of page

KITT AI Freight Optimizer

Created by: xenn0010 & Yeabsira Teshome


Freight loading inefficiency costs the logistics industry over. KITT tackles this massive problem with a physics-based AI world model inspired by cutting-edge research in autonomous agents.

What It Does

KITT predicts the optimal way to load trucks and containers — accounting for weight distribution, item geometry, weather, traffic, and route constraints.


KITT stood out for its ambition — a rare application of world models in an industry ready for AI-driven transformation.

How It Works

  • Leverages Redpanda for low-latency event streaming and agent coordination.

  • Predicts optimal positioning of every item inside a truck or container

  • Uses weather and traffic forecasts to adjust configurations dynamically

  • Aims to autonomously power the entire freight-loading workflow, end-to-end

Tech Stack

  • Backend: FastAPI, WebSockets, Python 3.10+

  • Streaming: Redpanda

  • Databases: SQLite, Neo4j

  • AI/ML: DeepPack3D, Claude 3.5 Haiku, scikit-learn

  • Voice: Pipecat AI

  • MCP: FastMCP for AI tool integration

  • APIs: OpenWeatherMap, TomTom Traffic, OpenRouteService

bottom of page