Owning Inference on a DGX Spark- Qwen3.6 27B writing code that ships
Detailed blog post about personal local LLM setup, not a reusable tool.
GLM-5.2 QuantTrio (Unpruned ) on 4xDGX Spark (GB10): 327K-655K context or up to 5 concurrent agents, one receipe. TP4 + DCP + MTP
GLM-5.2 deployment recipe for 4x DGX Spark with 655K context support.
ML engineers deploying large models on NVIDIA hardware
vLLM · TGI · SGLang
Detailed blog post about personal local LLM setup, not a reusable tool.
Generates clean, exportable Node.js workflows from prompts—skips SaaS lock-in, lands real code in git.
Outbound-only Pilot Agent beats Portainer for NAT traversal without exposing Docker socket.
Eight years of node ops expertise packaged into a self-hosted control panel.
Yet another self-hosting platform competing against Cloudron, CapRover, and YunoHost.
Modern monorepo replacing decade-old PHP restaurant platforms with 330+ tests and React Native.