Local diagnostic CLI for NVIDIA DGX Spark (GB10). Detects power caps, unified memory pressure, thermal risk, Docker/runtime issues, and validates vLLM/Ollama/llama.cpp/SGLang recipes.

Spark Pulse is a web control plane for spark-vllm-docker

Production-ready guide to deploy a high-performance local LLM server on NVIDIA DGX using llama.cpp, CUDA acceleration, and systemd automation.

MCP server for the Sovereign AI Blog: Search, retrieve, and diagnose tools for self-hosted AI on NVIDIA DGX Spark.

Local-first developer dashboard for the NVIDIA DGX Spark.

DGX Spark Local Inference Toolkit: llama.cpp, Vibe-Coding setup, and Whisper TUI.

3-bit Lloyd-Max KV Cache Compression for LLM Inference on NVIDIA DGX Spark GB10 — 5.12x compression, 0.983 cosine similarity, pure numpy on ARM unified memory

Deploy Nemotron 3 Nano 30B with 1M context window on NVIDIA DGX Spark using llama.cpp (Blackwell sm_121, Q4_0 KV cache quantization)

SageGPT (7.5M param SLM): A Transformer trained from scratch on ~140M pure Sanskrit tokens on NVIDIA DGX Spark. 6 Layer, 8 Attn Head, 256 embed, 1024 context, ~8K vocab.

Run Google Gemma 4 models efficiently on NVIDIA DGX systems using Hugging Face Transformers, CUDA acceleration, and optimized inference workflows. This repository provides a practical notebook for loading, configuring, and generating text with Gemma models on high-performance multi-GPU infrastructure.

Deploy Nemotron 3 Nano 30B on NVIDIA DGX Spark using TensorRT-LLM (Blackwell GB10, NVFP4 quantization, OpenAI-compatible API)

Complete self-hosted AI server stack on the Nvidia DGX Spark (arm64)