Artículo: AMZ-B0FT2S1D6T

Artificial Intelligence: AI Engineer's Cheatsheet: Silicon Edition (Ultra-large scale LLM training and inference)

Format:

Paperback

Kindle

Paperback

Detalles del producto
Disponibilidad
En stock
Peso con empaque
0.45 kg
Devolución
Condición
Nuevo
Producto de
Amazon
Viaja desde
USA

Sobre este producto
  • Ultimate guide to understand and design Large-scale LLM training and inference system for AI gigafactory era. "Artificial Intelligence: AI Engineer’s Cheatsheet - Silicon Edition" is a comprehensive technical companion for mastering AI systems engineering and large-scale LLM (Large Language Model) optimizations for the era of AI gigafactories. This book bridges the gap between high-level machine learning theory and low-level silicon-aware implementation - empowering engineers to design, train and deploy state-of-the-art (SOTA) models efficiently on modern hardware architectures. Unlike traditional ML/AI textbooks, this edition distills the knowledge required to build and optimize the same scale of systems powering organizations like OpenAI, Anthropic, DeepSeek and Google DeepMind. If you can understand the concepts in this book, you will not only understand how today’s AI systems work - you will be ready to work at leading AI labs. What You will Learn? This book systematically covers the essential layers of modern AI engineering:AI System Design (MLSys & Serving):End-to-end design concepts of AI serving systems such as vLLM, SGLang, and TensorRT-LLM. Explore scheduling, batching and optimizations like Continuous Batching - the same strategies enabling OpenAI to serve hundreds of millions of user queries weekly.Core LLM Architecture and Operations:A deep dive into Transformer-based architectures, including attention and their optimized variants such as FlashAttention, FlexAttention and memory-efficient decoding pipelines.Quantization Engineering:Understand BF16, FP8 (E4M3/E5M2) and quantization-aware training techniques for compute and memory optimization across GPUs, TPUs and custom accelerators.AI Hardware Architecture:A silicon-level exploration of GPUs and CPU backends (x86, ARM). Learn how hardware characteristics such as memory hierarchy and interconnect bandwidth impact model performance.Software-Hardware co-design:And much more. By the end of this book, you will be able to:Engineer and optimize end-to-end AI systems for large-scale training and inference.Evaluate trade-offs between model accuracy, latency, throughput and cost.Design quantization and parallelization strategies suitable for real deployments.Perform back-of-the-envelope calculations for compute, bandwidth, and memory requirements.Engage meaningfully in technical discussions on AI architecture, geopolitics and the emerging compute economy. Who This Book Is For?Students and developers preparing for machine learning, deep learning, and GenAI interviews.Engineers and researchers seeking to solidify their understanding of large-scale AI system design.Professionals transitioning into AI infrastructure, compiler or hardware optimization roles.Independent learners aiming to conduct research or replicate open-source SOTA systems.AI will not replace you - but engineers who understand AI systems from algorithm to silicon will. Start with this book, and redefine your position in the AI era. Book: Artificial Intelligence: AI Engineer's Cheatsheet: Silicon edition Author: Seymour Papermaster Updated: 1 November 2025 (v1.15) Pages: 206 Table of contents:Global AI race...Decode-Maximal BatchingContinuous BatchingP/D disaggregationCollective Communication primitivesMemory components in LLMTensor Parallelism [TP]...INT8 QuantizationGPU Workload Parallelizationand much more
$97,58
44% OFF
$54,21

IMPORT EASILY

By purchasing this product you can deduct VAT with your RUT number

$97,58
44% OFF
$54,21

3 meses de gracia en diferidos y hasta 6 meses sin intereses con Pacificard

Envío gratis
Llega en 5 a 12 días hábiles
Con envío
Tienes garantia de entrega
Este producto viaja de USA a tus manos en