I'm Wael Alzoubi — a full-stack & ML engineer who likes building hard things from scratch, every layer of the stack: neural nets in raw PyTorch, an operating system with its own C compiler, digital logic down to the gate, and production web apps in between.
Ordered by abstraction level — bottom of the stack to the top. Same engineer, all the way down.
A decoder-only transformer built from scratch in PyTorch — causal self-attention, pre-norm LayerNorm, GELU, and learned positional embeddings — that learns to generate Python source code character by character. No black-box libraries; the attention math is hand-written.
A tiny 32-bit operating system written in C — bootloader, kernel, filesystem, graphics, and syscalls — that boots on QEMU. It ships with its own in-OS C compiler (cc): write C inside the OS, compile it, and run the binary. Systems programming end to end.
Digital logic built up from first principles — from NAND/NOR gates to multiplexers, decoders, parallel adders, latches, and a working ALU with an instruction reader. The physical foundation everything above is standing on.
Production sites shipped for paying clients — designed, built, and hosted. All live right now.
Open to full-stack and ML engineering work. Always happy to talk shop — about attention heads, boot sectors, or anything in between.