Releases · AutomataNexus/Prometheus_Public

Prometheus v0.1.0

AI-powered edge ML training orchestrator with 6-stage pipeline: Ingest → Analyze → Train → Evaluate → Convert → Deploy.

13 neural network architectures — LSTM, GRU, RNN, Sentinel, Conv1d, Conv2d, ResNet, VGG, ViT, BERT, GPT-2, Nexus, Phantom
ONNX/HEF model conversion — export trained models for ONNX Runtime, TensorRT, OpenVINO, or Hailo-8/8L NPU
Full-stack Rust — Axum 0.7 server, Leptos 0.7 WASM UI, AxonML training, edge inference daemon
DigitalOcean Gradient AI agent — PrometheusForge analyzes datasets and recommends architectures via Claude with 24-document RAG knowledge base
CLI + TUI — Ratatui-based terminal UI with 8 tabs (Dashboard, Datasets, Models, Monitor, Agent, Convert, Quantize, Deploy)
Security — HashiCorp Vault secrets management, AES-256-GCM credential encryption, Argon2id password hashing, SSRF/SQLi guards
Stripe billing — 4 subscription tiers with usage-based token metering
Edge deployment — Push models to Raspberry Pi controllers via encrypted SSH. LSTM/GRU at 1.8 MB RSS, CNN via Hailo-8 NPU
Live sensor data — Real HVAC readings from Warren A-Wing AHU-6 controller every 10 seconds

Crate	Description
`prometheus-server`	Axum 0.7 HTTP server, REST API, WebSocket
`prometheus-ui`	Leptos 0.7 WASM SPA
`prometheus-training`	AxonML training pipeline (13 architectures)
`prometheus-shield`	Security layer — SSRF, SQLi, rate limiting, audit
`prometheus-edge`	Edge inference daemon (tiny_http)
`prometheus-reports`	PDF report generation
`prometheus-email`	Transactional email (Resend)
`prometheus-cli`	CLI + TUI client (Ratatui)
`prometheus-agent`	Python Gradient AI ADK agent

Built for the DigitalOcean Gradient AI Hackathon — March 2026
By Andrew Jewell Sr. — AutomataNexus