01Track 01 · Systems Foundations
The substrate everything else runs on.
Before you can reason about ML compilers or GPU kernels, you need to understand what the CPU is actually doing on every clock cycle. This track unlearns the abstractions that high-level languages provide and rebuilds your mental model around hardware reality.
- — stack vs heap, RAII, move semantics, smart pointers
- — cache lines, branch prediction, data-oriented design
- — threads, lock-free programming, vector intrinsics
- Read C++ code and predict where it allocates, where it copies, and where it stalls
- Recognize and fix cache-unfriendly memory access patterns
- Reason about why a “trivial” change to a loop can be 10× faster
- Have the vocabulary to read a CPU profiler output without getting lost