About the author
Curiosity, made executable.
I’m Akiwayne. My starting point is Go backend development. My current focus is the engineering around AI systems—and understanding what happens inside them.
What I work through
I’m interested in the boundary between reliable services and model inference: how requests enter a system, how work is bounded, how cancellation propagates, and how we measure the result. Python and PyTorch offer another view of those questions, closer to tensors and the computation itself.
How I write
These notes start with an implementation or a question I can test. I explain the mechanism, show where a simple approach breaks, and record the assumptions that make a result meaningful. A local experiment is described as a local experiment.
The companion labs are intentionally small. You should be able to run them, change an input, and discover where an explanation stops being sufficient.
Start here
Read the Engineering Reliable Model Systems series, or explore the projects and their validation boundaries.