Senthilkumar Gopal

This post analyzes Hashcash as a defense against AI web crawlers, detailing its probabilistic proof-of-work model and operational viability. It includes a technical breakdown of token generation, expected runtime guarantees, and implementation strategy.

This post explains how Paged Attention and Chunked Prefill optimize memory and computation in vLLM by organizing key-value caches into dynamic blocks and processing input sequences in manageable chunks. It includes a simple walkthrough with tensor shapes and code to showcase their integration for LLM inference.

This post covers practical Slurm commands and best practices to help you reserve, access, and work cleanly on nodes.

A quick clarification between the terms - Triton, TensorRT, and TensorRT-LLM

A runbook for triaging Neuron accuracy issues and means to verify the model accuracy

This post works through a high level overview of OpenMPI and EFA

A growing list/study guide of AI compilers, from foundational concepts like graph lowering and systolic arrays to practical tools like TorchDynamo and Glow.

This post explores the concept of aliasing in XLA, its significance, the mechanisms through which it is implemented, and future directions for extending aliasing optimizations.

This post works through the metrics that are critical for ML inference

This post works through the metrics that are critical for ML inference