PyTorch
Building the framework up from its one real primitive — the tensor — through autograd and the training loop.
Engineering
5 parts · Sep 08, 2026 – Sep 11, 2026 · 29 min
Some subjects took more than one post. These are the runs, in the order they were written — every part published, none of them a cliffhanger.
2 series, 8 parts in total.
Building the framework up from its one real primitive — the tensor — through autograd and the training loop.
5 parts · Sep 08, 2026 – Sep 11, 2026 · 29 min

What MCP solves, what happens to the context window at scale, and where it sits next to A2A.
3 parts · Feb 22, 2025 – Feb 25, 2025 · 12 min