SYNQAGI StateDelta Edge

Local-first Physical AI that detects expected-vs-observed state deviations on Arm edge devices, combines deterministic safety rules with advisory AI, and keeps sensitive data offline.

Comment

posted an update —

SYNQAGI StateDelta Edge — Local AI Runtime Milestone

SYNQAGI StateDelta Edge has progressed from its deterministic state-conformance core to a working local AI advisory runtime.

During the challenge period I:

  • added a local Ollama advisory runtime adapter
  • preserved deterministic StateDelta safety validation as the final operational authority
  • constrained model reasoning to validated delta/evidence and allowed actions
  • added protection against invented evidence paths
  • created a reproducible llama3.1:8b local inference baseline
  • expanded the automated test suite to 13 passing tests
  • validated CI across Python 3.10, 3.11, and 3.12
  • published the work in the public Apache-2.0 repository

The next milestone is the Local Decision Orchestrator, followed by profiling and optimization for NVIDIA Jetson Orin on Arm64.

Built for the Physical AI track: local-first, expected-vs-observed state intelligence with deterministic safety boundaries.

PhysicalAI #EdgeAI #Arm64 #LocalAI #AIOptimization

Log in or sign up for Devpost to join the conversation.