My AMD GPU Has a New Job: Running llama.cpp Without Guessing What Fits
How I run llama.cpp on an AMD GPU with HIP, verified GGUF weights, a private API and measured context limits instead of optimistic VRAM maths.
Karl Machleidt
Curiosity.
Trial. Error.
Security research, self-hosting, and the things I learn by taking things apart.
Read the blogLatest stories
Ghostwire: Reading My Own WhatsApp Desktop Without Clicking Through It
How Ghostwire uses a native Windows WebView bridge for local exports and authenticated webhooks, with session, compatibility and queue limits.
I Gave My AI Agents Credentials. I Would Rather Not Give Them All of Them.
How I use self-hosted Infisical, machine identities and selected credential injection for AI tools, with reproducible examples and honest permission boundaries.
I Built a Search Setup That Gives My Local Model Something to Read
Vane, SearXNG and local inference: searchable sources, honest engine failures, document conversion and a reproducible way to check the answers.
Pre-Astro: How I Got My WordPress Cache Above 90%
The caching setup I used before Astro: anonymous HTML at Cloudflare, an origin page cache, Redis limits and tests for private-content bypass.
My Websites Have a Bouncer Now: OPNsense, NAXSI and a Very Particular Block Button
A technical tour of my OPNsense nginx reverse proxy, NAXSI policies, Elastic evidence and guarded n8n response, including tuning, tests and current limitations.
CitrixPot: My Low-Interaction NetScaler Honeypot and What Actually Hit It
Why I built a low-interaction NetScaler honeypot, how its capture-only stack works, and what real payloads and the private evidence console actually show.
I Took WordPress Off the Internet: Karl.Fail on Astro and Cloudflare Pages
How I moved Karl.Fail to Astro and Cloudflare Pages, kept WordPress private, and built automatic publishing, local search and a faster reading…
