eugene-plexus

Retired: conversation-history storage from the former Eugene Plexus consciousness framework. Historical code, not part of the current inference control plane.

Retired: training execution and checkpoints for the former Eugene Plexus LLM training platform. Not part of the current inference control plane.

Retired: pipeline coordination for the former Eugene Plexus LLM training platform. Historical code, not part of the current inference control plane.

Retired: model-serving component from the former Eugene Plexus LLM training platform. Current inference uses upstream engines with agent, gateway, and inference-driver.

Deferred: historical Discord/chat-platform bridge from the former Eugene Plexus consciousness framework. Not yet integrated with the current inference control plane.

Retired: dataset preparation and tokenization for the former Eugene Plexus LLM training platform. Not part of the current inference control plane.

Model library for Eugene Plexus: GGUF and safetensors discovery, metadata, launch profiles, resumable downloads, and hardware fit guidance. Your files stay yours.

Per-host node agent for Eugene Plexus: supervises components and inference engines, manages engine binaries, and handles runtime admission and node enrollment.

One HTTP adapter per model backend for Eugene Plexus: fronts OpenAI-compatible engines, hosted APIs, and subscription CLIs.

OpenAI-compatible gateway for Eugene Plexus: routes local and cloud model backends with replica balancing, priority-tier failover, and runtime lifecycle policy.

Retired: constitution, self-model, and person records from the former Eugene Plexus consciousness framework. Not part of the current inference control plane.

Retired: evaluation suites and checkpoint comparison for the former Eugene Plexus LLM training platform. Not part of the current inference control plane.

Eugene Plexus control root: install-wide trust, node enrollment, topology, and a single-writer replicated log with warm standbys. Spawns no processes.