ML Systems Notes

Static docs mirror of JINO-ROHIT/ml-systems-notes — a personal collection of notes on ML systems engineering covering distributed computing, parallelism, quantization, and pytorch internals.

Use the top nav or search to explore. Python examples are inlined below each section's README.

Sections

• Distributed Techniques — NCCL collectives, MoE, parallelism strategies, torch.distributed
• Quantization — symmetric/asymmetric, LLM.int8(), AWQ, SmoothQuant, GPTQ/OBS/OBQ, QuIP
• Torch Notes — autograd, dynamo, AOTAutograd, FX graphs, inductor, torch.compile
• SGLang — KV cache internals
• CUDA Graphs
• Compute — LLM FLOPs
• JAX Scaling Book — roofline analysis