This repository contains the initial (V1) local AI orchestration engine for GemmaEdge, originally developed as a hackathon project. It served as a Proof of Concept (PoC) to demonstrate how to run a multi-agent AI reasoning system on strictly constrained consumer hardware (GTX 1050 Ti), completely bypassing cloud dependencies and API costs.
As my personal development standards and system architectures have evolved, this specific codebase is now considered legacy and no longer reflects my current operational level.
While the code itself is outdated, the foundational system design and the Reasoning over Reflex logic remain structurally sound. If you want to dive into the core architecture, the MishaSelector routing mechanics, and the VRAM cold-swap logic, please refer to the deep-dive documentation:
👉 Read the system design and logic here: ARCHITECT.md
GemmaEdge V2 is currently in active development. The new iteration leaves the V1 hackathon constraints behind, featuring a cleaner, highly modular architecture that elevates local "Sovereign" AI orchestration to the next level.
If you are looking for my active production code or want to see my current capabilities, please check the other repositories on my GitHub profile.