First, split the design into two parts: the skeleton and the “brain” part. Use 2×4s for the backbone—those are the ones that actually hold the weight, so pick a consistent color for the frame so you can see the load paths. Then use smaller bricks like 1×2s, 1×4s, and 1×1s for the neural‑network detail. If you want to show weights, put a little sticker or a different colored tiny piece on the top of each 1×2 to indicate weight magnitude. When you stack layers, keep the heavy bricks on the bottom and the lighter ones on top. Think of each layer like a “block” in a real neural net: the lower layers carry the bulk of the data (or bricks). Add a bit of a 2×4 cross‑beam every two layers to keep the sides from swaying. Prototyping: build one layer, load it with a weight, and see if it bends. If it does, replace the 2×4 with a 2×6 or add a diagonal brace. Keep a spare stash of 2×4s “just in case” because you’ll want to swap out anything that feels weak. And remember—glue is great, but don’t over‑apply it or your bricks will get soggy. A quick dry‑fit first, then add glue for the final push. That way you can tweak the structure before it’s set. Happy building!