Building a BGP Daemon in Rust
From wire codec to 22 interop suites in six weeks. Architecture decisions, performance wins, and what happens when your test dependency drops the feature you need.
#rust#bgp#networking#rpki#aspa
blog
Near-Lossless Inference at 183 tok/s: Gemma 4 MoE on RTX 5090
How Gemma 4's sliding window architecture lets you run Q8_0 on a 32GB card. 183 tok/s generation with near-lossless quality at 200K context.
#gemma-4#llm#inference#rtx-5090#llama-cpp#rookery#moe
The Single-GPU Inference Stack in 2026
Running a 35B model at 213 tokens per second on a single RTX 5090, managed by an open-source Rust daemon with crash recovery, hot-swap, and agent lifecycle.
#rust#llm#inference#rtx-5090#llama-cpp#rookery