Nils Jannasch
I help enterprises adopt AI without losing control of it. Senior Solutions Engineer at Wiz focused on AI & code security — before that, a decade of pre-sales and platform engineering at Palo Alto Networks and Bosch. Off the clock I hold myself to one rule: never write about AI systems I haven't run, benchmarked, or broken first. That discipline became a long-running series from vibecoded applications through local inference to agent sandboxing — every claim measured, every setup reproducible.
Start here: RTX 5090 AI Box on Proxmox — 160K context, 150 t/s local Qwen
Currently tinkering with
📡ESP32 WiFi radar with probe request sniffing 🧪TurboQuant KV cache: 400K context on 16GB VRAM 🖨️Multi-agent 3D model generation pipeline
Explore by Topic
🧠
Local AI & Inference
Running large models on consumer hardware. Benchmarks, configs, and what actually works.
18 posts Latest: RTX 5090 AI Box on a GMKtec Proxmox Host: LXC, 160K Context, and 150 t/s Local QwenRecent Posts
All posts → RTX 5090 AI Box on a GMKtec Proxmox Host: LXC, 160K Context, and 150 t/s Local Qwen Qwen 3.8 27B on a 5060 Ti: 64 t/s Local Coding and a Car Configurator That Actually Works Beryl AX OpenWrt Mobile Office Router with Antigravity CLI Sandboxing AI Agents: Kernel-Level Isolation with nono and Landlock From Vibe Coding to AI Agent: My Local Qwen 3.6 Now Runs 24/7
"The Lovecraft of AI" — Victor, Coworker