Large language models

You tried AI tools. They're fast. And you're still doing all the work — just in a chat window.

Building a Native LLM Inference Engine Building a modern LLM inference engine from the ground up requires understanding every layer between model weights and generated tokens. This book provides a systems-level, implementation-focused path through that entire stack, from raw checkpoint formats and tensor memory layouts to native C++ execution, CUDA kernels, KV-cache management, scheduling, multi-GPU inference and production serving.

For PMs, presales, and managers who work with AI but don’t write code: how RAG, agents, fine-tuning, compliance, and security actually fit together.

The model is one box. This book is about every other box: retries, contracts, evals, budgets, guardrails and agents for backend engineers shipping LLM features to production.

What if the next leap in AI is not just predicting the next word, but learning how to reason? Reasoning Models are changing how AI approaches complex problems, from mathematics and coding to multi-step decision-making. This beginner-friendly book explores the ideas behind this shift, including Chain-of-Thought (CoT), Test-Time Compute, Reinforcement Learning, and the connection between reasoning models, Generative AI, and AI Agents. Discover what makes reasoning AI different, why it matters, and how it may shape the future of artificial intelligence — without getting lost in complex mathematics or dense technical details.

Mastering LLM Quantization: Architecture and Code A Comprehensive Engineering Guide to Low-Precision AI, Quantized Inference, GPU Kernels, and Production Deployment Mastering LLM Quantization: Architecture and Code is a comprehensive technical resource for understanding, implementing, optimizing, and deploying quantized large language models. Combining mathematical foundations with practical engineering techniques, this title explores how reduced-precision representations can significantly improve memory efficiency, inference performance, and hardware utilization while preserving model accuracy.

Você viu um modelo de linguagem fazer meia hora de trabalho sozinho. Este livro trata da pergunta que veio logo depois: até que ponto você pode confiar nele?

Has visto a un modelo de lenguaje hacer media hora de trabajo por su cuenta. Este libro trata de la pregunta que vino después: ¿hasta qué punto puedes confiar en él?

From Mathematical Foundations to the Future of AIMost deep learning still lives in flat, Euclidean space. Real-world data does not.This 15-chapter university textbook takes you from the basics of topology and geometry to the frontier of Geometric Deep Learning, Topological Data Analysis, equivariant networks, and optimization on manifolds.Clear explanations. Worked examples. Research roadmaps.

Skip the trial and error. 53 agent skills picked and run in Claude Code and Codex, each with who it fits, install commands, and a first run.

What if a 125B-parameter AI model could run locally on a gaming GPU with just 12 GB of VRAM? Strata makes it possible. This book takes you inside the techniques that make it work, from MoE sparsity and memory tiers to PCIe pipelining and speculative decoding, then shows you how to install, tune and troubleshoot it yourself.

Most people use AI. Few understand how far they can take it. AI Levels gives you a practical roadmap from prompting and everyday AI use to workflows, agents, specialized systems, and AI building—one skill at a time.

La inteligencia artificial está cambiando las reglas. AI Levels te lleva, paso a paso, de usuario de IA a creador de soluciones inteligentes: aprende, crea, automatiza y construye con IA.

사내 LLM 게이트웨이를 설계하고 운영하는 실무서입니다. 키리스 인증과 6단계 RBAC, 토큰 쿼터·예산과 비용 통제, 엔터프라이즈 규모 MCP, 운영 런북까지 다룹니다. 영문판 The AI Gateway Playbook의 한국어판(PDF 75쪽)입니다.

A practitioner's playbook for building an in-house LLM gateway: keyless auth, six-tier RBAC, quotas and budgets, per-token cost control, MCP at enterprise scale, and the operations to run it all. About 32,000 words with diagrams, configs, and checklists.