Notebooks
This is where Braid publishes its substantive writing: engineering studies, technical posts, and white papers. We publish here when we have something worth coming back to.
JAX under the hood: optimization tricks and profiling tips.
A tour of JAX's execution pipeline — tracing, jaxprs, and XLA — and how understanding the internals lets you write faster code and read a profiler with confidence.
Taking derivatives with respect to shapes, à la Hadamard.
How you take a derivative with respect to a shape — Hadamard's 1908 construction, worked through two concrete examples, and why it underpins shape optimisation for real engineering problems.
The hybrid approach to AI-powered discovery.
Why AURA combines symbolic reasoning with learned components, and what each contributes to the system's reliability on real engineering problems.
A note. Notebooks is not a feed and not a blog. We add to it when there is a piece worth writing and reading. If you would like to be told when something new appears, the most reliable way is to come back here periodically.