Large language models

Master DeepSeek V3 — one of the most capable AI models of its time.This practical guide teaches students, researchers, and professionals how to use DeepSeek V3 for learning, research, coding, content creation, automation, and productivity — with strong focus on prompt engineering and responsible AI usage.

Unlock the power of Llama 4 — the next generation of open-weight multimodal AI.This practical guide shows students, researchers, and professionals how to harness advanced AI tools for learning, research, teaching, and productivity. From generating study notes and research ideas to building educational chatbots and optimizing workflows, discover how Llama 4 can transform the way you work and learn

A visual journey through the world of Large Language Models. Learn how modern AI systems work using diagrams, mental models and simplified explanations—without getting lost in unnecessary complexity

AI application development is moving fast, but building systems that actually work in production is still difficult. This guide shows you how to build, deploy and scale AI applications with Dify, covering visual workflows, RAG, AI agents, integrations, security and production operations. It's a practical, hands-on resource for turning prototypes into reliable real-world applications.

From Prompt to Autonomous Agent is a practitioner’s guide to building AI systems that actually run in production – written for engineers who build things that need to work at 3am, six months after launch.The book teaches twelve levels of agentic AI architecture using a single, unbroken running example: a job failure in AWS Glue that requires a root cause diagnosis. This same system evolves over the fifteen chapters - from a single API call in Chapter 1 to a Society of Agents deliberating over a novel multi-factor failure in Chapter 12. You don’t just know what each pattern is by the end. You know what it is for, how much it costs, and what it gives you over the level below.Each chapter focuses on a single pattern: the Basic Responder, the Router, Tool Calling, Memory Augmentation, the Autonomous Agent, Reflective Self-Critique, Human in the Loop, Planning, Multi-Agent Coordination, Tool Building, Hierarchical Orchestration, and the Society of Agents. Each chapter has working code (one file, under 100 lines, no dependencies other than the Anthropic SDK) and a Failure Museum - a catalogue of real production failures with the signal, the cause, and the fix.The final section, Production and Craft, covers what most AI books entirely skip over, how to make agentic systems observable, how to protect them with cost guardrails, how to build fallback chains that degrade gracefully, and how to evaluate them with enough discipline that you know they're working before your users tell you they aren't.The book is for data engineers, backend engineers, and ML engineers who are done with demos and ready to ship. Requires Python fluency and engineering judgement. No prior knowledge of AI research, math or language models is assumed.The examples are in Python and use the Anthropic SDK because that’s where real systems live. They are based on real AWS infrastructure , EventBridge, Glue, Lambda, Bedrock, Snowflake, OpenSearch.The core principle runs through all the chapters: the right level is always the lowest level that can reliably handle your problem. This book will show you how to figure out what level that is.

When something trips and two hundred alarms fire at once, which one is the cause? From Flood to Cause shows you how to build deterministic, grounded, auditable root-cause analysis where a causal engine decides and a language model only explains — never diagnoses, never guesses, and abstains the moment grounding is absent. Every concept is wired to a live, zero-install NEXUS-1 console in your browser, so you don't just read the architecture — you drive it.

The hardest part of building AI is not the model. It is everything around it. The Art of Harness Engineering is a practical guide to turning AI prototypes into reliable products. It covers testing, guardrails, observability and governance, giving you the tools to build AI systems people can trust and organizations can run with confidence.

Most developers use AI coding tools like better autocomplete. The 10x Developer's Cursor Playbook shows you how to use Cursor for agent workflows, refactoring, MCP integrations, collaboration and security so you can ship faster, write better code and get more out of AI.

The EU AI Act is messy, confusing, and unavoidable. Get compliant before it becomes a problem.

Vibe coding makes it easier than ever to build software fast, but shipping a real product takes more than speed. Vibe Coding to Production shows how to turn AI-generated code into secure, reliable and scalable SaaS applications with practical advice, real-world examples and the engineering habits needed to build production-ready systems.

AI is changing exploit development. Learn how to use large language models with tools like IDA Pro, Ghidra and Binary Ninja to speed up vulnerability research, reverse engineering and proof-of-concept exploit development while understanding where human expertise still matters.

As AI adoption accelerates, attacks such as prompt injection, jailbreaks, data poisoning, and agent exploitation are redefining cybersecurity. This book explains why these attacks work and provides practical strategies for building secure, resilient AI systems.

From LLMs and RAG to Memory, MCP, Agent Frameworks and Enterprise AI Controls—discover how modern AI Agents work internally within today's rapidly evolving AI ecosystem.

Most AI systems can talk, but few can actually do. This book shows you how to build AI agents that reliably use tools, call APIs and automate real workflows. Using DSPy, Pydantic AI, the Claude Agent SDK, the OpenAI Agents SDK and Google ADK, you'll learn practical patterns for building reliable agents that work in production.

There are no magic prompts to test AI. What you can do is give AI real work under realistic conditions and observe what happens - exactly the kind of evaluation now required under the EU AI Act.