Reading
A practical tutorial on LangChain and LangGraph for developers, covering complete workflows such as RAG systems, intelligent search agents, and text classifiers, providing directly runnable Python scripts instead of Jupyter notebooks
Section 01
This course is a practical tutorial for AI application developers, presented in pure Python script form. It covers complete content such as LangChain basic components, LangGraph workflow orchestration, and advanced RAG systems, helping developers quickly master the core capabilities of both frameworks and cultivate engineering development habits.
Section 02
The LangChain LangGraph Crash Course (LLCC) is positioned as a practical tutorial for mastering the frameworks from scratch. Unlike the Jupyter Notebooks in official documentation, it uses fully annotated pure Python scripts that can be run directly. The learning path is: first master LangChain basic components, then orchestrate complete workflows via LangGraph. It is suitable for engineers with Python basics who want to quickly get started with large model application development.
Section 03
The first example simple_llm_application.py demonstrates the basic LLM calling pattern, building a chat application via Chat Models and Prompt Templates. It emphasizes the declarative chain assembly syntax of LangChain Expression Language (LCEL) to improve readability and extensibility.
The second example semantic_search_engine.py introduces basic RAG components, demonstrating content loading from PDFs, vector embedding (text-embedding-3-large), and vector database storage. It reveals the advantage of semantic search in capturing semantic relationships through vector similarity.
text_classifier.py uses large model structured output to implement text classification; data_extractor.py extracts structured data via Few-Shot Prompting, demonstrating LangChain's ability to build text understanding pipelines.
Section 04
agent.py builds an intelligent agent with web search capabilities, defining tools, system prompts, and decision loops, and autonomously decides whether to call tools to integrate external information. LangGraph provides clearer architectural abstraction and state management capabilities.
agent_human_assistance.py implements the Human-in-the-loop mode, where the agent pauses to wait for human input when encountering uncertain situations, ensuring controllability and safety in high-risk scenarios.
agent_time_travel.py demonstrates state history management functions, supporting viewing execution trajectories, backtracking to any state, and modifying decisions to re-execute, helping debug complex agent behaviors.
Section 05
rag.py implements the standard RAG process (document loading → chunking → embedding → indexing → retrieval → generation), introducing Self-Query technology that allows large models to automatically generate retrieval filters to improve accuracy (e.g., filtering by extracted year and topic).
rag_delegation.py uses the ReAct mode (Reasoning + Acting), where the agent alternates between thinking (analyzing information, planning actions) and acting (retrieval/computation), enhancing the ability to handle complex multi-hop problems.
Section 06
Advanced resources recommended by the course:
Version reminder: The tutorial is based on pre-1.0 versions of LangChain. The core logic remains unchanged, but a few syntax adjustments are needed by referring to the migration guide.
Section 07
Learning suggestions: Practice step by step in the course order—first master the independent use of basic components, then learn complex workflow orchestration. The pure Python script form is more in line with engineering code organization methods, cultivating good development habits.
Summary: LLCC focuses on practicality, quickly building hands-on experience through runnable code. It is a high-quality resource for building applications after understanding the principles of large models.
Keep going with more reads from the same topic.
Splinter is a minimalist, high-performance key-value (KV) and vector storage system enabling zero-latency inter-process communication via shared memory and atomic operations. With only 766 lines of core code, it supports millions of operations per second and 768-dimensional vector storage, offering a new architectural approach for local LLM inference and data-intensive applications.
Folkering OS is the world's first AI-native bare-metal operating system, entirely written in Rust no_std without relying on Linux, POSIX, or libc. It can generate commands from scratch, compile them into WASM, and run them in 10 seconds, achieving true self-evolution.
This project explores how to use Large Language Models (LLMs) to detect logical vulnerabilities in Ethereum smart contracts, especially business logic flaws that are difficult for traditional static analysis tools to capture, such as economic manipulation and execution order errors.
This article introduces an open-source project for building modern large language models from scratch. The project evolves from a GPT-2-style basic architecture to a Llama 2/3-style production-level implementation, covering complete tutorial implementations of key technologies such as RMSNorm, RoPE, SwiGLU, GQA, and MoE.