30 PyTorch problems cover MLSys inference, quantization, KV cache, and batching

kalyan_kpl · x · 2026-07-23

A 30-problem PyTorch sheet for learning MLSys and inference, covering attention variants, MoE, quantization, KV cache, decoding, batching, and distributed inference systems.

The sheet is organized into sections and is meant as a hands-on study aid rather than a tutorial blog post. The accompanying image shows the full curriculum, including attention architectures, efficient transformer components, serving schedulers, and distributed inference topics.

Original post →

More from coding & agent

coding & agent channel →