Stage 4 · Mastery & Industry
Compute & Hardware
GPUs, memory bandwidth, clusters and the chips that power AI.
3 lessons
- 01 GPUs, Memory and FLOPs: Why AI Needs Them Matrix maths, memory bandwidth and why inference is often memory-bound. Intermediate 15 min ✎ In progress
- 02 Distributed Training and Clusters Data, tensor and pipeline parallelism, interconnects and failures at scale. Expert 20 min ✎ In progress
- 03 The AI Chip Landscape GPUs, TPUs, custom accelerators and export controls. Advanced 15 min ✎ In progress
Lessons here are still being written — start with these course units
Each unit has goals, algorithms, must-read papers, a project and a self-test.