SeshaOS

NexusAOS v2 — governance-first, local-first AI OS with specialist local models (Gemma 4, Qwen3) and LiteLLM-compatible routing.

Overview

SeshaOS is the evolution of NexusAOS, designed as a governance-first, local-first AI OS. It uses specialist local models and maintains offline-first execution with reversible, permissioned actions.

Specialist Models

Gemma 4 12B serves as the Planner, Qwen3-Coder 30B handles Implementation, and Qwen3.5 9B manages Vision tasks. This separation of concerns improves reliability and performance.

Governance & Routing

Kernel-centric governance where models propose actions and the kernel validates/records every state change. LiteLLM-compatible proxy support enables seamless integration with NVIDIA NIM and other providers.

Tech Stack

RustLiteLLMGemmaQwenLocal LLMsGovernanceEvent SourcingUbuntu