LangChain & LLMs - Build Autonomous AI Tools Masterclass

Set up your development environment with Python and VSCode, obtain an OpenAI API key, and securely manage credentials and billing while exploring the API playground.

Learn to install Python across Windows, Mac, and Linux by following a detailed online guide, verifying the installation with version commands, and optional customization.

Install and configure Visual Studio Code with the Python extension pack and environment manager to streamline Python development for building autonomous AI tools with the OpenAI API.

Explore what a large language model is, how it’s trained on vast textual data, and how transformers with self-attention generate natural, human-like text.

Explore the LangChain ecosystem from core packages and the LangChain expression language to Langraph, Langsmith, and Langserve, plus integration packages.

Discover LangChain v1 architecture, from application layer to runnables, LCEL, and model integrations, and learn how prompts, models, and chains become runnables via pipe and unified invoke, batch, and stream.

Explore LangChain building blocks: chains, agents, and modules, through hands-on installation, Python environments, and OpenAI integration, with documentation guiding retrieval chains, memory, and callbacks.

Explore how large language models are trained and how Lang Chain functions as a framework, then dive into the building blocks of Lang Chain.

Explore chat prompt templates and multi-message prompts, including system, human, and AI messages. Learn to build modular, reusable prompt components and ground LLMs for robust AI tools.

Hands-on exploration of LangChain v1 prompt templates and messages, using system and human prompts, message placeholders, and few-shot techniques to enable dynamic, reusable prompt design and prompt composition.

Explore output parsers hands-on by building simple chains using string, json, and pedantic output parsers, and transition to structured output to extract typed data from LLM responses.

discover how to use langchain output parsers, including the pydantic output parser, to declare a python data model with a json schema and parse lms data with validation.

Learn to build robust data extraction with Pydantic and a pedantic parser to define a vacation data model, validate fields, and extract structured trip details from emails.

Explore structured outputs with output parsers, including string, json, and pydantic parsers, and build nested schemas like address and company using the modern structured output method and line chain v1.

Explore LangChain concepts, including lcl and runnable chains, prompts, chat prompt templates, and output parsers, and learn to build a basic chain with a prompt, model, parser, and pipe operator.

Build and run basic and parallel chains using LangChain with chat models, prompts, and parsers, then aggregate results from summary, keywords, and sentiment analyses in parallel.

Explore how to build a pass-through chain using runnable parallel, a fake retriever, and prompt templates to pass context, question, and model output.

Explore chain branching by building a conditional routing in LangChain chains using a classifier prompt and runnable branches, then test code vs general prompts with two LLM calls per request.

Explore debugging chains in LangChain and LLM apps by inspecting prompt, model, and output parser, examining input and output schemas, using config tracing, and logging steps to verify chain interfaces.

Explore memory in LangChain using the modern message history approach within a session store, comparing buffer, window, and summary strategies, and learn to persist chat history with SQLite.

Create and manage a conversation memory in LangChain by building a memory.py workflow that stores chat history and enables the model to recall details like user name and topics.

Manage multiple conversation sessions by building per-user memories within a shared chain of prompts, messages, and history, enabling each user’s memory and session memory to reference past chats.

Learn how to trim messages to fit a context window by applying token-aware strategies, preserving system messages and choosing last or first messages to reach a target token count.

Master windowed memory by implementing a sliding window that keeps only the last two exchanges, balancing token costs with lost context.

Build a chat bot with persistent memory using SQLite, integrate sql chat message history, and automate summarization after 10 messages to persist across restarts.

Learn how LangChain document loaders convert raw files like pdf, text, html, and csv into document objects with content and metadata. Compare pypdfloader, textloader, web-based loader, and unstructured loader.

Explore document splitting with LangChain and LLMs, using pypdfloader to load a pdf and a recursive character text splitter to create chunked documents with metadata.