Deep dives on systems I find worth fully understanding — networking fundamentals, ML inference, and the internals of tools I use every day.
A complete reference on the OSI model — all seven layers, encapsulation, ARP, TCP vs UDP, and the routing protocols (RIP, OSPF, BGP) that stitch the internet together. Ends with why all of this matters for distributed ML systems.
Read postTwo notes in one: a summary of the NVIDIA CUDA C++ Programming Guide (kernels, thread hierarchy, memory hierarchy, heterogeneous execution) followed by a walkthrough of PyTorch internals — tensors, strides, the two-level dispatch chain, autograd, and what writing a kernel actually looks like.
Read post →