Blog

Technical writing on systems, AI, and engineering from the collective.

Sep 8, 2026 · Alish Karki · 5 min read

Building PRISM: I spent months writing an LLM inference engine in C99

I built PRISM, a plugin-based CPU inference engine from scratch. This is the real story — the bugs, the benchmarks, and why I stopped.

c
llm
inference
systems

Sep 4, 2026 · Alish Karki · 4 min read

Building vek: A Zero-Dependency SIMD Vector Similarity Library in C

How I built vek, a hand-tuned SIMD vector similarity library with zero dependencies. Covers CPU dispatch, Neumaier summation, and masked AVX-512 loads.

simd
vector-search
c
performance