Description
Unleash the power of AI agents on Arm-based hardware! In this hackathon category, your mission is to develop an innovative, AI-powered agent that runs efficiently on an Arm-based system. We encourage creative solutions that demonstrate practical applications of AI agents, regardless of complexity.
Your agent should leverage a Large Language Model (LLM) – either an API-based model (like OpenAI API, Gemini API, etc.) or an on-device LLM running via llama.cpp or a similar framework. While API-based models offer flexibility, utilizing on-device LLMs is highly encouraged to maximize performance and demonstrate the full potential on Arm architecture. We encourage participants to design their agents with safety and prompt guardrails in mind, aiming to prevent unintended consequences.
Key Requirements
- Arm-Native: Your AI agent, including its LLM inference (if on-device), must run on an Arm-based system. This could be anything from a mobile device (like a Google Pixel phone), a single-board computer (like a Raspberry Pi), a laptop (such as an Apple Silicon Mac), or a cloud server running on Arm architecture.
- Clear Use Case: The agent should demonstrate a clear, even if simple, practical application.
- Efficiency: Consider resource usage and performance optimization on your chosen Arm device.
- Safety Considerations: Design with basic safety measures and prompt guardrails in mind.
Prize
The winning team will receive a Raspberry Pi Zero 2 W kit to continue their AI agent exploration!
