31 repos. All Rust. All public. All live. Every badge is a shipped product with a timestamp proving I'm still building it.
Loading badges...
Timestamps refresh every 30 minutes via GitHub API. Yes, I cache. No, I don't fake the numbers.
The Stack
Language: Rust — 100% of production code
No Python in production. No JavaScript frameworks. No Go. No Java. One language, compiled to native machine code, across every product.
13 years of defense and enterprise. One language. Zero runtime dependencies.
Infrastructure: 4 bare metal Debian nodes + Mac Mini
lf (20 cores, 750GB, RTX 3070 8GB VRAM) · gd (20 cores, 760GB, RTX 3050 Ti 4GB VRAM) · bt (12 cores, 95GB) · st (14 cores, 767GB)
Mac Mini ARM for development. All connected via SSH mesh. Static IPs on 192.168.1.50-79 sled.
GPU nodes power MoE model training, Pixel Forge diffusion models, and on-device inference. Laptop GPUs — not datacenter cards — proving you don't need $10K hardware to run AI locally.
$10/month total. Zero cloud. Zero containers. Zero Kubernetes.
AI: signed-weights inference, on-device only
No cloud inference in production. Every model compiled into the binary passes NanoSign verification at load. Custom diffusion models for image generation. LoRA fine-tuning. Signed speech-to-text. All on-device, all verified, no unsigned weights in the production path.
AI without the cloud bill. AI without the data leak. AI without the unsigned weights.
Certifications & Registrations
SDVOSB (submitted) · SAM.gov (active) · Maryland eMMA (vendor) · CSB (approved)
Army 17C Cyber Operations · JCAC 2014 · USCYBERCOM J38
Every badge earned. Every registration filed. Every repo proves the claim.